Microwave Extraction Unlocks Quercetin and Saponins from Discarded Banana Corms

A part of the banana plant that farmers usually throw away is emerging as a surprisingly rich source of medicinal chemistry. In a study published in Food Chemistry: X, researchers in Vietnam report that the corm of Musa balbisiana, a wild banana relative common across Southeast Asia, can yield extracts packed with polyphenols and saponins when processed with microwave-assisted extraction. Using a carefully optimized set of conditions, the team recovered more than 71 milligrams of polyphenols, expressed as gallic acid equivalents, per gram of dry corm material, alongside roughly 61 milligrams of saponins per gram. Spectroscopic analysis of the purified extract then pointed to quercetin, one of the best-studied antioxidant flavonoids in nature, as a principal component. The work is notable not only for what it extracted but for how rigorously it tuned the process, applying statistical design methods to balance two chemically very different classes of compounds in a single green-chemistry protocol.

The significance of the target material lies in its status as an agricultural by-product. Musa balbisiana is a wild progenitor of many cultivated banana varieties, and its corm, the underground bulb-like stem from which the pseudostem grows, has long featured in folk medicine across Vietnam, Laos, and Thailand, where preparations have been used against stomach ulcers, kidney stones, metabolic disorders, and diabetes. Earlier laboratory studies have supported some of these traditional claims: corm extracts have shown antioxidant activity, hypoglycemic effects in vitro and in vivo, and inhibition of metabolic enzymes such as tyrosinase, xanthine oxidase, and lipase, which are implicated in skin disorders, gout, and obesity respectively. Yet despite this pharmacological promise, corms are routinely discarded after harvest, representing a wasted stream of raw material. Turning that waste into standardized, bioactive extracts would fit squarely within the growing circular-economy push in food and pharmaceutical research, where by-products are reimagined as feedstocks for nutraceuticals, cosmetics, and functional foods.

The extraction technology at the heart of the study, microwave-assisted extraction, works on a principle quite different from conventional soaking or Soxhlet methods. Microwaves heat the solvent and the plant matrix from within, generating localized thermal gradients that rupture cell walls and drive dissolved compounds rapidly into the surrounding liquid. The result is dramatically shorter processing times and lower solvent consumption, both important credentials for a technique marketed as green. But microwave extraction is not a one-size-fits-all affair. The optimal combination of solvent composition, power, irradiation pattern, and duration depends heavily on the specific plant tissue and the chemistry of the target molecules, which is why the Vietnamese team insisted that conditions must be tailored to each new matrix. Previous studies had applied the technique to banana peels and other Musa waste streams, recovering phenolics with water or glycerol-ethanol mixtures, but no published protocol existed for extracting both polyphenols and saponins from M. balbisiana corm.

The researchers began with a systematic one-factor-at-a-time exploration of the key variables. Fresh corms collected in Dong Thap province were cleaned, sliced, dried at 60 degrees Celsius, ground into a fine powder, and extracted with aqueous methanol at ratios of one gram of material to thirty milliliters of solvent. Polyphenol recovery climbed steadily as methanol concentration rose from 40 to 60 percent, reaching a plateau above that level, while saponin content followed the same trajectory, peaking near 70 percent methanol. The explanation is a tug-of-war between polarity and microwave physics: moderately polar phenolic aglycones and glycosylated triterpenoid saponins dissolve better in less polar mixtures, but above roughly 60 to 70 percent methanol the solvent’s dielectric properties deteriorate, so it can no longer convert microwave energy into heat efficiently. Power told a similar story of diminishing returns, with 360 watts delivering the best yields but no statistically significant gain over 270 watts, and higher settings risking thermal degradation of heat-sensitive compounds.

Irradiation cycling and extraction time completed the picture. Pulsing the microwaves for four seconds per minute outperformed both shorter and longer duty cycles, an effect the authors attribute to the need for rest intervals that allow active compounds to diffuse from the plant matrix into the solvent between heating pulses. Extending the total extraction time from 20 to 40 minutes significantly boosted yields, with polyphenols reaching 71.63 milligrams per gram and saponins 60.45 milligrams per gram, but longer exposures caused yields to slip, consistent with cumulative microwave-induced degradation. Screening experiments using a two-level factorial design then confirmed which variables truly mattered. Microwave irradiation cycle, power, and solvent concentration emerged as the three dominant factors at the 5 percent significance level, while extraction time proved the least influential and was fixed at 40 minutes for all subsequent modeling.

With the influential variables identified, the team turned to response surface methodology using a Box-Behnken design, a statistical framework that fits a quadratic model to experimental data and locates the optimum without testing every possible combination. Fifteen experimental runs spanning three levels of methanol concentration, power, and irradiation cycle produced regression equations for both target responses. The models performed well, explaining 94 percent of the variance in polyphenol yield and 95 percent in saponin yield, with non-significant lack-of-fit tests indicating that the fitted surfaces faithfully represented the underlying chemistry. Strikingly, the response surfaces for the two compound classes were nearly congruent, both showing dome-shaped profiles with maxima near the center of the experimental domain. This overlap means a single set of operating conditions serves both chemically distinct metabolite families, a practical advantage for any future industrial process aiming at a dual-action extract.

The predicted optimum called for 70 percent methanol, 270 watts of microwave power, and a 4-second-per-minute irradiation cycle. Validation experiments under these conditions delivered 71.35 plus or minus 0.51 milligrams of polyphenols and 60.68 plus or minus 1.23 milligrams of saponins per gram of dry matter, deviations below 5 percent from the model predictions and statistically indistinguishable from them. The extract was then concentrated and fractionated through liquid-liquid partitioning with n-butanol and water, followed by silica gel chromatography eluted with chloroform-methanol gradients. Thin-layer chromatography, visualized with ferric chloride for phenolics and anisaldehyde-sulfuric acid for saponins, guided the collection of enriched fractions for structural analysis.

The spectroscopic characterization formed the analytical centerpiece of the work. Raman spectra displayed intense bands for conjugated aromatic carbon-carbon stretching between 1600 and 1620 wavenumbers, along with ring deformation modes and sugar-region signals near 843, 974, and 1100 wavenumbers, features compatible with glycosylated flavonoids and carbohydrate-bearing constituents. The authors were careful to note, however, that these Raman features are supportive rather than confirmatory for saponins, since no band uniquely identifies a triterpenoid or steroidal saponin skeleton. Fourier-transform infrared spectroscopy told a complementary story, revealing a broad hydroxyl stretch near 3400 wavenumbers, a conjugated carbonyl at 1666 wavenumbers characteristic of the gamma-pyrone ring of flavonoids, and multiple aromatic ring stretches between 1450 and 1611 wavenumbers, together with carbon-oxygen bands in the fingerprint region consistent with glycosidic moieties.

Nuclear magnetic resonance provided the most specific identification. The purified fraction, designated compound CC, showed five aromatic proton signals in its proton NMR spectrum, including the meta-coupled pair at 6.18 and 6.40 parts per million diagnostic of the A-ring of a flavonol, and the three-proton pattern of a catechol-type B-ring. A strongly deshielded singlet at 12.49 parts per million confirmed the 5-hydroxyl group hydrogen-bonded to the C-4 carbonyl, while carbon NMR displayed fifteen signals including a carbonyl at 176.3 parts per million. Every chemical shift matched published quercetin values within 0.3 parts per million for carbon and 0.15 for hydrogen. The authors nonetheless label the identification tentative: without high-resolution mass spectrometry or two-dimensional NMR experiments, the molecular formula could not be verified experimentally, and follow-up analyses are planned. No individual saponin was isolated at all, with evidence for that class resting on colorimetric assays, TLC staining, and vibrational bands. Those caveats aside, the study makes a persuasive case that a discarded banana corm, zapped with carefully tuned microwaves, can deliver pharmaceutical-grade chemistry from the compost heap.

Subject of Research: Microwave-assisted extraction and spectroscopic characterization of polyphenol- and saponin-rich extracts from Musa balbisiana corm

Article Title: Extraction and spectroscopic characterization of polyphenol- and saponin-rich extracts from musa balbisiana corm with tentative identification of quercetin

Article References: Nhon, H. T. N., Trang, N. T. H., & Anh, L. T. H. (2026). Extraction and spectroscopic characterization of polyphenol- and saponin-rich extracts from musa balbisiana corm with tentative identification of quercetin. Food Chemistry: X, Article 104578. https://doi.org/10.1016/j.fochx.2026.104578

Image Credits: AI Generated

DOI: 10.1016/j.fochx.2026.104578

Keywords: Musa balbisiana, banana corm, microwave-assisted extraction, polyphenols, saponins, quercetin, response surface methodology, Box-Behnken design, FT-IR spectroscopy, Raman spectroscopy, NMR, agricultural by-products