An integrated green strategy based on deep eutectic solvents and ultrasound for efficient polyphenol profiling and antioxidant evaluation of dendrobium officinale Authors: Dong-min Cao, Qian Zeng, Wen-xing Ao, Shuai Zheng, Ying-xin Lu, Xi-cheng Wang, Ming-Hua Xian, Shu-mei Wang Published: 2026 ABSTRACT Deep eutectic solvents (DESs) combined with ultrasound-assisted extraction (UAE) have emerged as a promising green strategy for efficient recovery of bioactive compounds. In this study, a DES-based UAE system was developed for the extraction of polyphenols from Dendrobium officinale (DO), with particular emphasis on the synergistic effects and underlying mechanisms. Among 24 screened DESs, choline chloride-fructose (ChCl-Fru, 1:2) exhibited the highest extraction efficiency. Key extraction parameters were optimized using Plackett-Burman design and Box-Behnken design, yielding an optimal total polyphenol content (TPC) of 87.75 mg GAE/g DO. Kinetic modeling revealed that the extraction process followed a second-order model, indicating that both mass transfer and solute–solvent interactions governed the extraction behavior. Compared with silent conditions, ultrasound significantly enhanced extraction efficiency. The TPC obtained by DES-UAE (88.11 mg GAE/g DO) increased by 130% relative to silent DES extraction, which was notably higher than that observed in the 70% methanol system (107%). In addition, DES-UAE extracts exhibited superior antioxidant activities, with DPPH and ABTS scavenging rates of 90.79% and 96.46%, respectively. Molecular dynamics simulations further demonstrated that the DES system provided a more favorable solvation environment for phenolic compounds, characterized by enhanced hydrogen-bonding interactions, reduced molecular aggregation, and stronger binding energies compared with methanol. These findings reveal that the superior performance of DES-UAE originates from the synergistic interplay between ultrasonic cavitation and the unique physicochemical properties of DES. Chemical profiling by UHPLC-Q Exactive Orbitrap HRMS identified 31 phenolic compounds, including flavonoids, phenolic acids, biphenyls, and other derivatives. Multivariate statistical analysis revealed significant metabolic differences among DO samples from different geographical origins, and 11 key differential metabolites were identified as potential biomarkers. In addition, the DES-UAE extracts exhibited superior antioxidant activity compared to conventional ethanol extracts. Overall, this study provides a mechanistic understanding of DES-UAE systems and offers a green and efficient strategy for the extraction and utilization of polyphenols from DO. Introduction Widely distributed across medicinal and edible plants, polyphenols constitute a class of bioactive compounds endowed with antioxidant, anti-inflammatory, and anti-aging effects [1]. Given these multifaceted bioactivities, polyphenol-rich natural products have garnered increasing interest in recent years for their potential applications in functional foods, pharmaceuticals, and nutraceuticals [2]. Among these, Dendrobium officinale (DO) stands out as a particularly valuable source, renowned for its high nutritional and medicinal value [3]. Previous studies [4] identified a range of bioactive constituents in DO, including polysaccharides, flavonoids, phenolic acids, and bibenzyl compounds, which contribute to its antioxidant and health-promoting properties. Beyond antioxidant activity, Dendrobium officinale has also been reported to exert multiple biological functions [5]. Recent studies [6] have demonstrated its potential in modulating gut microbiota composition and alleviating intestinal barrier damage, particularly under pathological conditions such as diabetes. These findings further highlight the broader health significance of DO and underscore the importance of efficiently extracting its bioactive constituents. Given the richness of its polyphenolic constituents, developing efficient methods for their extraction from the plant matrix is critical to fully exploiting the potential of this valuable resource. Traditional approaches to polyphenol extraction from plant matrices commonly employ organic solvents such as methanol, ethanol, and acetone [7]. Although these solvents are effective in extracting phenolic compounds, their use often raises concerns regarding toxicity, environmental impact, and solvent residue [8]. In addition, traditional extraction techniques generally require long extraction times and large volumes of organic solvents, which limit their sustainability and industrial applicability. Therefore, the development of environmentally friendly and efficient extraction technologies for plant polyphenols has become an important research topic. DESs have emerged as a promising class of green solvents in recent years. DESs are typically composed of a hydrogen bond acceptor (HBA) and a hydrogen bond donor (HBD), with strong hydrogen bonding between the components yielding a eutectic mixture whose melting temperature is significantly depressed relative to the neat compounds [9]. Compared with conventional organic solvents, DESs possess several attractive properties, including low toxicity, biodegradability, easy preparation, and tunable physicochemical characteristics [10]. These advantages make DESs highly suitable for the extraction of natural products from plant materials. For instance, DESs have demonstrated up to a 1.34-fold higher yield of blueberry polyphenols compared to 70% ethanol under similar conditions [11]. Meanwhile, ultrasound-assisted extraction (UAE) has been widely recognized as an efficient technique for enhancing mass transfer during extraction processes. The cavitation effect generated by ultrasound can disrupt plant cell walls, facilitate solvent penetration, and accelerate the release of intracellular compounds, thereby significantly improving extraction efficiency [12]. Although the combination of DESs and ultrasound has attracted significant attention as a promising candidate for the extraction of bioactive compounds, several key issues remain insufficiently understood [13]. Previous research has centered on the optimization of extraction conditions, while the molecular-level interaction mechanisms between DESs and target compounds have rarely been investigated [14]. In particular, the role of hydrogen bonding and intermolecular interactions in promoting the solubilization of polyphenols in DES systems remains largely unclear. Molecular dynamics simulation has emerged as a powerful computational tool for exploring intermolecular interactions and solvation behavior at the molecular level, providing valuable insights into the mechanisms underlying solvent–solute interactions [15]. Therefore, the objective of this study was to develop a green and efficient DES-UAE method for the extraction of polyphenols from DO. Optimization of the extraction process was carried out using Plackett-Burman (PB) design in combination with Box-Behnken response surface methodology. The extraction kinetics were further investigated to elucidate the mass transfer behavior during the DES-UAE process. In addition, the chemical composition of the extracted polyphenols was characterized using UHPLC-Orbitrap high-resolution mass spectrometry. Importantly, molecular dynamics simulations (MDS) were employed to investigate the intermolecular interactions between DES components and representative polyphenolic compounds, aiming to provide molecular-level insights into the enhanced extraction performance of the DES-UAE system. DPPH• and ABTS+ radical scavenging assays were utilized to determine the antioxidant activities of the extracts. This study provides a comprehensive understanding of the synergistic effects of DESs and ultrasound in the extraction of plant polyphenols and offers theoretical guidance for the need for environmentally friendly extraction approaches for natural products. Methods and materials Materials This study’s source of DO is detailed in Table S1. The plant materials were ground and sieved through an 80-mesh screen to achieve a uniform particle size. All chemicals, including choline chloride (ChCl), glucose, fructose, glycerol, propylene glycol, malic acid, lactic acid, and acetylacetic acid, were of analytical grade or higher and were purchased from McLin Chemical Reagents Co., Ltd. Gallic acid was obtained from Chengdu Kroma Technology Co., Ltd. Synthesis of DESs 24 DESs were prepared by simple heating and stirring their constituent components at a set molar ratio [16]. Briefly, a homogeneous DES was successfully formed by heating a mixture of HBD and HBA at 80 °C with stirring for 1.5 h, with no crystallization or precipitation observed after 24 h of cooling and standing. The chemical compositions, molar ratios and abbreviations of the DESs are summarized in Table S2, with the characterization are provided in Supplementary methods. Determination of total polyphenol content After 30-fold dilution, the extract was subjected to a modified Folin-Ciocalteu spectrophotometric assay to obtain its total polyphenol content (TPC) [17]. Following the addition of 20 µL of Folin-Ciocalteu reagent to 20 µL of extract and 5 min incubation, 160 µL of 10% Na2CO3 solution was added. After 60 min reaction at room temperature, the absorbance was measured at 765 nm using a Synergy HTX Multimode Reader (BioTek Instruments, USA). Gallic acid was used as the standard, and TPC was expressed as milligrams of gallic acid equivalent per gram of dried Dendrobium officinale (mg GAE/g DO). Antioxidant capacity The antioxidant activity of DO polyphenols was evaluated using DPPH• and ABTS+ radical scavenging assays. The experimental procedures are as follows: First, the polyphenol extract was diluted to different concentrations as needed. In the DPPH• radical scavenging assay, 20 µL of the sample solution was mixed with 180 µL of 0.1 mmol/L DPPH• solution and reacted in the dark for 30 min. The absorbance was measured at 517 nm. In the ABTS+ radical scavenging assay, 100 µL of the sample solution was mixed with 100 µL of ABTS+ working solution and reacted in the dark at room temperature for 10 min. The absorbance was measured at 734 nm. Optimization of extraction experiments Selection of optimal DESs for ultrasound-assisted polyphenol extraction 24 DESs were sequentially examined via ultrasound-assisted extraction to pinpoint the most efficient DES for isolating polyphenols from DO. All extractions were carried out under uniform conditions to ensure consistency. Briefly, 0.1 g of DO powder was added to 3.0 mL of the prepared DES (40% aqueous solution) in a 5 mL tube. The mixture was transferred to a thermostatically controlled ultrasonic bath (KQ-300DE, Kunshan ultrasonic Instruments Co., Ltd., China) operating at a frequency of 40 kHz and a nominal power of 210 W, with the actual calorimetric power determined as 50.5 W. Extraction was performed at 40 °C for 30 min. Following extraction, the homogenate was centrifuged at 13,000 rpm for 15 min, and the clear supernatant was retained for subsequent analysis. In parallel, control extractions were performed using water, ethanol/water (70:30, v/v) and ethanol as the solvent under the same experimental conditions. Total polyphenol extraction efficiency was quantified spectrophotometrically with the Folin-Ciocalteu reagent, enabling a direct comparison of the extraction performance among the tested DESs. The DES affording the highest polyphenol yield was designated as the optimal medium and adopted for all subsequent investigations The actual ultrasonic power was determined calorimetrically [18]. Briefly, the bath was filled with 7 L of water and operated for 30 min at nominal powers of 150–270 W. Temperatures were measured at fifteen positions in triplicate. The actual powers measured for nominal settings of 150, 180, 210, 240, and 270 W were 36.1, 43.3, 50.5, 57.8, and 65.0 W, respectively. To ensure accuracy, all subsequent analyses employed these measured actual power values. Single-factor experiments Extraction performance in the DES-UAE system is governed primarily by the solvent formulation and the operational parameters employed. A systematic set of single-factor assays was performed to delineate how individual variables modulate the TPC of DO extracts. These factors include the water content of ChCl-fructose DES (20%, 30%, 40%, 50%, and 60%), liquid-to-solid ratio (20, 30, 40, 50, and 60 mL/g), ultrasonic power (36.1, 43.3, 50.5, 57.8, and 65.0 W), ultrasonic temperature (30, 40, 50, 60, and 70℃), and ultrasonic time (10, 20, 30, 40, and 50 min). PB experiments PB design was employed to identify the critical variables governing polyphenol recovery from DO among five candidates: ultrasonic time, ultrasonic power, ultrasonic temperature, DES water content, and liquid-to-solid ratio (Table S3). The standardized effect normal plot was used to assess the position of effect points relative to the reference line, while the standardized effect Pareto chart was used to evaluate the significance of main effects or interaction effects based on the size of the columns [18]. Following identification of factors with significant effects on the TPC of DO, response surface methodology (RSM) was employed for optimization. Box-Behnken design (BBD) experiments Building upon single-factor and PB results, a BBD was subsequently constructed with TPC (mg GAE/g DO) as the dependent response. Employing the factor-level combinations embedded in the design matrix (Table S4), the predicted optimum was located and subsequently corroborated by the verification trial to attest the reliability of the fitted model. Comparison with silent extraction conditions To evaluate the contribution of ultrasound to the extraction efficiency, control experiments were conducted under silent (non-ultrasound) conditions. Extractions were performed using the optimal DES (ChCl-Fru, 1:2) and 70% ethanol as solvents under identical conditions of power (50.5 W), temperature (50 °C), time (39 min), liquid-to-solid ratio (54 mL/g), and water content (28.6% for DES), but without application of ultrasound. Instead, the mixtures were maintained under magnetic stirring at 500 rpm. All other procedures were consistent with those described in Sections 2.5.1. TPC and antioxidant activities were compared between ultrasound-assisted and silent extraction to quantify the synergistic effect of ultrasound. Extraction kinetic model This study investigated the extraction kinetics of TPC from DO under the conditions optimized using RSM for DES-UAE and 70% ethanol extraction. Both first-order and second-order kinetic models were applied, and the suitability of each model in describing the extraction process was compared. MDS analysis MDS were performed using the GROMACS 2025.3 software package with the general AMBER force field (GAFF) [19]. The molecular structures of 2-coumaric acid, tristin, quercetin, methanol, choline chloride, and fructose were obtained from the PubChem database. Their topologies and initial configurations were generated based on the GAFF force field in combination with the TIP3P water model, with water molecules added to fill the remaining space. Prior to the production run, energy minimization was carried out using the steepest descent algorithm to eliminate unfavorable atomic contacts. Subsequently, the system was equilibrated under NVT and NPT ensembles to ensure stable temperature and pressure conditions. During the simulations, all bonds involving hydrogen atoms were constrained using the LINCS algorithm. The temperature was maintained at 298.15 K using the V-rescale thermostat, while the pressure was controlled at 1.0 bar using the Parrinello-Rahman barostat. Long-range electrostatic interactions were treated using the particle mesh Ewald (PME) method, and a cutoff distance of 1.0 nm was applied for van der Waals interactions. The integration time step was set to 2 fs, with a total simulation time of 100 ns, and trajectory snapshots were recorded every 1.0 ps for subsequent analysis. After the simulations, trajectory analyses were performed using built-in GROMACS tools to evaluate structural stability, molecular flexibility, and intermolecular interactions. Visualization and graphical representations were generated using VMD, Origin, and BIOVIA Discovery Studio Visualizer 2019. Identification of phenolic compounds in Dendrobium by UPLC-MS Phenolic profiling was performed on a Dionex UltiMate 3000 UHPLC system hyphenated to a Q-Exactive Orbitrap high-resolution mass spectrometer (Thermo Fisher Scientific, MA, USA). An Acquity UPLC BEH C18 column (2.1 mm × 100 mm, 1.7 µm) held at 40 °C was employed for chromatographic resolution using 0.1% aqueous formic acid (solvent A) and acetonitrile (solvent B) as the mobile phase. A 24-min gradient was programmed at 0.3 mL/min: 5% B (0–2 min), ramped to 50% B (2–4 min), then to 60% B (4–11 min) and 90% B (11–20 min), held at 90% B for 1 min (20–21 min), returned to 5% B within 1 min (21–22 min), and re-equilibrated at 5% B for 2 min (22–24 min). High-resolution MS data were acquired in positive and negative heated electrospray ionisation (HESI) modes with the following settings: sheath gas 45 (arbitrary units), auxiliary gas 10(arbitrary units), spray voltages +3.2 kV/−2.5 kV, capillary temperature 300 °C and auxiliary gas heater temperature 100 °C. Statistical analysis All experiments were performed in triplicate, and results are presented as mean ± standard deviation. One-way analysis of variance (ANOVA) and Duncan’s multiple comparison test were conducted using SPSS 23.0 software. Design Expert 13.0 was utilized for the design and data analysis of ramp-up and response surface experiments. Multivariate statistical analyses encompassing principal component analysis (PCA), hierarchical clustering analysis (HCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were performed with SIMCA 14.1. Meanwhile, GraphPad Prism 9 served as the platform for additional data analysis and scientific graph plotting. Results and discussion Screening of DESs for polyphenol extraction 24 DESs composed of choline chloride as HBA and various HBDs were evaluated for their extraction efficiency toward polyphenols from DO (Fig. 1A). Extraction performance varied significantly across the DES systems. Among them, the DES composed of choline chloride and fructose (ChCl-Fru, 1:2, DES-8) exhibited the highest extraction yield of total polyphenols, achieving a maximum TPC of 24.10 ± 0.37 mg GAE/g DO (Fig. 1B). This result is consistent with the study by Yiwei Zhu [16], who used eight DES solvents to extract polyphenols from walnut shellsFig. 1Appearance of Dendrobium officinale polyphenols extract (A). TPC of Dendrobium officinale extracts obtained by 24 DESs and 3 conventional solvents coupled with UAE (B). FT-IR spectra of DES-8, choline chloride, and fructose (C). Effect of different water content on TPC (D), liquid–solid ratio (E), ultrasonic time (F), ultrasonic power (G), ultrasonic temperature. Chcl-Glu, Choline chloride-Glucose; Chcl-Gly, Choline chloride-Glycerol; Chcl-Fru, Choline chloride-Fructose; Chcl-Egly, Choline chloride-Ethylene glycol; Chcl-Mac, Choline chloride-Malic acid; Chcl-Lac, Choline chloride-Lactic acid; Chcl-Ure, Choline chloride-Urea; Chcl-Aac, Choline chloride-Acetylacetic acid. Appearance of Dendrobium officinale polyphenols extract (A). TPC of Dendrobium officinale extracts obtained by 24 DESs and 3 conventional solvents coupled with UAE (B). FT-IR spectra of DES-8, choline chloride, and fructose (C). Effect of different water content on TPC (D), liquid–solid ratio (E), ultrasonic time (F), ultrasonic power (G), ultrasonic temperature. Chcl-Glu, Choline chloride-Glucose; Chcl-Gly, Choline chloride-Glycerol; Chcl-Fru, Choline chloride-Fructose; Chcl-Egly, Choline chloride-Ethylene glycol; Chcl-Mac, Choline chloride-Malic acid; Chcl-Lac, Choline chloride-Lactic acid; Chcl-Ure, Choline chloride-Urea; Chcl-Aac, Choline chloride-Acetylacetic acid. To further elucidate the interactions between HBA and HBD during DES formation, FT-IR spectra of DES-8, choline chloride, and fructose were analyzed. As shown in Fig. 1C, all three samples exhibited broad absorption bands in the O-H stretching region (3685–3053 cm−1), which are characteristic of hydrogen-bonded hydroxyl groups. Compared with the individual components, DES-8 displayed a more intense and broadened peak centered at 3284 cm−1, indicating the formation of stronger hydrogen-bonding interactions during DES formation. Similar characteristic peaks were observed in the C–H stretching region (3049–2838 cm−1), the C–H bending vibration region (1524–1402 cm−1), and the C–C stretching vibration region (1368–1278 cm−1), suggesting that the fundamental structural features of the components were retained in the DES system. In addition, DES-8 showed N-H bending vibrations in the range of 1716–1586 cm−1, similar those of choline chloride, further confirming that no new covalent bonds were formed during DES synthesis. Notably, DES-8 exhibited evident red shifts in the regions of 1115–987 cm−1 and 987–903 cm−1, corresponding to the stretching vibrations of C–O and C–N bonds, respectively. These spectral shifts indicate the formation of strong intermolecular hydrogen bonding interactions between choline chloride and fructose, which contribute to the establishment of a stable hydrogen-bonding network within the DES system. Such a hydrogen-bonding network is considered a key factor responsible for the superior extraction performance of DES-8. The multiple hydroxyl groups in fructose can effectively interact with the phenolic hydroxyl groups of polyphenolic compounds, thereby enhancing their solubility in the DES matrix. In addition, the physicochemical properties of the DES, including viscosity and mild acidity, may further facilitate the dissolution of polyphenols [20]. Additionally, the mild acidity of DES-8 further contributes to the extraction efficiency. Therefore, the extraction efficiency of DES systems depends on both the structural features of the plant matrix and the solvent’s physicochemical properties, including viscosity and pH (Table S2). Based on these findings, DES-8 was chosen as the optimal solvent for subsequent optimization. Effect of DES-UAE extraction parameters on TPC To evaluate the influence of extraction parameters on DES-UAE process, single-factor experiments were conducted by varying water content in DES, liquid-to-solid ratio, ultrasonic time, ultrasonic power, and ultrasonic temperature. The results are presented in Fig. 1D–H. Water content serves as a critical factor in modulating the physicochemical characteristics of DES, particularly viscosity and polarity. As shown in Fig. 1D, the TPC initially increased with increasing water content and reached a maximum value of 30.61 ± 0.72 mg GAE/g DO at 30% water content. The addition of an appropriate amount of water reduces the viscosity of DES, thereby enhancing solvent diffusivity and facilitating mass transfer within the plant matrix [21]. However, excessive water weakened the hydrogen-bonding interactions between the HBA and HBD components, thereby disrupting the supramolecular network structure of the DES. This structural disintegration compromises the integrity of the hydrogen-bonding network and alters the solvent polarity, leading to a collapse of the tailored solvation environment that is critical for stabilizing phenolic compounds, ultimately reducing their solubility [22]. The liquid-to-solid ratio is another important parameter affecting extraction efficiency. As illustrated in Fig. 1E, the TPC increased significantly as the liquid-to-solid ratio increased from 20 to 50 mL/g. A higher solvent volume enhances the contact between the solvent and plant material, thereby increasing the concentration gradient and promoting the diffusion of polyphenols into the solvent phase. This can be explained by the enhanced movement space for DO powder, allowing the solvent to come into better contact with the material surface, thereby improving TPC extraction efficiency [23]. When the ratio further increased to 60 mL/g, a slight decrease in TPC was observed, which may be attributed to excessive dilution and reduced cavitation efficiency in the extraction system [24]. Ultrasonic parameters also exerted a significant influence on extraction performance. As shown in Fig. 1F, the TPC increased markedly with ultrasonic time from 10 to 40 min due to the intensified cavitation effect, which promotes the disruption of plant cell walls and facilitates the release of intracellular polyphenols into the solvent [25]. However, prolonged ultrasonication (50 min) led to a decrease in TPC, likely due to the degradation of phenolic compounds [26]. Similarly, the TPC increased with ultrasonic power from 36.1 to 50.5 W (Fig. 1G), indicating enhanced cavitation intensity and improved mass transfer [27]. Nevertheless, further increasing the power to 65.0 W resulted in a noticeable decrease in extraction efficiency, possibly due to localized overheating and degradation of polyphenols. Temperature also influenced the DES-UAE extraction process (Fig. 1H). The TPC exhibited a slight increase as temperature rose, culminating at 50 °C, which can be ascribed to enhanced molecular diffusion and mass transfer [28]. However, further increases in temperature resulted in a slight decline in TPC. This phenomenon may be associated with the reduced surface tension of the solvent at higher temperatures, which weakens cavitation intensity during ultrasonication and may also accelerate the degradation of phenolic compounds [29]. Based on the above results, the optimal extraction conditions were determined as follows: water content of 30%, liquid-to-solid ratio of 50 mL/g, ultrasonic time of 40 min, ultrasonic power of 210 W, and ultrasonic temperature of 50 °C. These parameters were subsequently used for further process optimization. PB design for screening significant factors affecting TPC A PB design was employed to screen the significant factors influencing the extraction of polyphenols from DO. Five variables were evaluated, including ultrasonic time, ultrasonic power, ultrasonic temperature, water content in DES, and liquid-to-solid ratio. The PB design enables rapid identification of the most influential factors while minimizing the number of experimental runs. The analysis results are presented in Table 2. With a p-value of 0.0002, the model was statistically significant, demonstrating that the selected variables had a significant impact on TPC. In addition, the high coefficient of determination (R2 = 0.9663) suggested that the model explained 96.63% of the variability in the TPC, demonstrating a satisfactory goodness of fit. Based on the experimental data summarized in Table 1 and Table S3, Pareto charts were generated to visualize the relative significance of each factor (Fig. 2A and B). The results revealed that ultrasonic time, water content in DES, and liquid-to-solid ratio were the most significant parameters affecting the DES-UAE extraction efficiency of polyphenols from DO. Notably, water content exhibited a negative effect on the extraction yield. Therefore, subsequent RSM experiments will focus on optimizing these three factors: ultrasonic time, DES water content, and liquid-to-solid ratio.Table 1ANOVA and significance test of regression models for PB design experiments.SourceSum of SquaresdfMean SquareF-valuep-valuesignificanceModel639.25915127.851834.37340.0002significantA-X1109.02501109.025929.31200.001significantB-X25.577015.57741.49950.2666C-X35.903415.90341.58710.2545D-X4167.28401167.284544.97500.0005significantE-X5351.46701351.467994.4932<0.0001significantResidual22.317063.7195Cor Total661.576111C.V.%4.770R20.9663Note: df: Degrees of freedom; Model: X1, ultrasonic time; X2, ultrasonic power; X3, ultrasonic temperature; X4, water content; X5, liquid-to-solid ratio; Cor Total: Corrected average of the sum of all information; C.V. %: Coefficient of variation.Fig. 2The normal plot (A) and the pareto chart (B) obtained from plackett-burman experiments showing the significance of the primary and interaction effects. Factor X1, ultrasonic time; Factor X2, ultrasonic power; Factor X3, ultrasonic temperature; Factor X4, water content in DES; Factor X5, liquid–solid ratio. Table 1ANOVA and significance test of regression models for PB design experiments.SourceSum of SquaresdfMean SquareF-valuep-valuesignificanceModel639.25915127.851834.37340.0002significantA-X1109.02501109.025929.31200.001significantB-X25.577015.57741.49950.2666C-X35.903415.90341.58710.2545D-X4167.28401167.284544.97500.0005significantE-X5351.46701351.467994.4932<0.0001significantResidual22.317063.7195Cor Total661.576111C.V.%4.770R20.9663Note: df: Degrees of freedom; Model: X1, ultrasonic time; X2, ultrasonic power; X3, ultrasonic temperature; X4, water content; X5, liquid-to-solid ratio; Cor Total: Corrected average of the sum of all information; C.V. %: Coefficient of variation. ANOVA and significance test of regression models for PB design experiments. Note: df: Degrees of freedom; Model: X1, ultrasonic time; X2, ultrasonic power; X3, ultrasonic temperature; X4, water content; X5, liquid-to-solid ratio; Cor Total: Corrected average of the sum of all information; C.V. %: Coefficient of variation. The normal plot (A) and the pareto chart (B) obtained from plackett-burman experiments showing the significance of the primary and interaction effects. Factor X1, ultrasonic time; Factor X2, ultrasonic power; Factor X3, ultrasonic temperature; Factor X4, water content in DES; Factor X5, liquid–solid ratio. Optimization of DES-UAE extraction conditions using BBD A BBD experiment was applied to further optimize the extraction parameters affecting the recovery of polyphenols from DO. Following the PB screening, ultrasonic time (A), DES water content (B), and liquid-to-solid ratio (C), which were selected for response surface optimization. The results of the BBD are summarized in Table S4, where the TPC values ranged from 64.76 to 86.91 mg GAE/g DO. The ANOVA results were presented in Table 2. The regression model was highly significant with an F-value of 308.39 (p < 0.0001), while the lack-of-fit was not significant (p > 0.05), indicating that the model adequately described the experimental data. In addition, the high coefficient of determination (R2 = 0.986) demonstrated that the model explained more than 98% of the variability in TPC. The low coefficient of variation (CV = 0.72%) further confirmed the good precision and reliability of the model. The following predictive model equation for TPC was derived through regression analysis:(7)YTPC=86.65+0.0497×A-2.03×B+5.64×C+3.08×AB-0.4599×AC+0.947×BC-4.72×A2-6.68×B2-7.19×C2Table 2ANOVA and significance validation for the response surface model of TPC yield.SourceSum of SquaresdfMean SquareF-valuep-valueSignificanceModel865.7969996.1997308.3886<0.0001significantA-X10.019810.01980.063370.8085B-X233.0333133.0333105.8954<0.0001significantC-X3254.83831254.8383816.9388<0.0001significantAB38.0183138.0183121.8757<0.0001significantAC0.846110.84612.712510.1436BC3.587113.587111.49910.0116significantA276.7388176.7388246.0028<0.0001significantB2187.67241187.6724601.6239<0.0001significantC2217.75281217.7528698.0532<0.0001significantResidual2.183670.3119Lack of Fit1.717730.57264.91640.0790Not significantPure Error0.465940.1165Cor Total867.9816Std.Dev.0.5585R20.9975Adj. R20.9942Pred. R20.9675Mean78.12C.V.%0.7150 Table 2ANOVA and significance validation for the response surface model of TPC yield.SourceSum of SquaresdfMean SquareF-valuep-valueSignificanceModel865.7969996.1997308.3886<0.0001significantA-X10.019810.01980.063370.8085B-X233.0333133.0333105.8954<0.0001significantC-X3254.83831254.8383816.9388<0.0001significantAB38.0183138.0183121.8757<0.0001significantAC0.846110.84612.712510.1436BC3.587113.587111.49910.0116significantA276.7388176.7388246.0028<0.0001significantB2187.67241187.6724601.6239<0.0001significantC2217.75281217.7528698.0532<0.0001significantResidual2.183670.3119Lack of Fit1.717730.57264.91640.0790Not significantPure Error0.465940.1165Cor Total867.9816Std.Dev.0.5585R20.9975Adj. R20.9942Pred. R20.9675Mean78.12C.V.%0.7150 ANOVA and significance validation for the response surface model of TPC yield. To visualize the interaction effects of the variables on TPC, three-dimensional response surface plots and corresponding contour plots were generated (Fig. 3A–F). The nearly circular contour plots observed in Fig. 3A and D indicate that the interaction between ultrasonic time and liquid-to-solid ratio was relatively weak, suggesting that these two variables influenced the extraction process independently. In contrast, the steeper response surface observed for liquid-to-solid ratio indicates that this factor exerted a stronger influence on TPC compared with DES water content. Overall, the increase in TPC with increasing liquid-to-solid ratio can be attributed to the enhanced contact between the solvent and plant matrix, which improves the dissolution and diffusion of target compounds [30]. Meanwhile, a moderate amount of water in the DES system reduces solvent viscosity and improves polarity, thereby facilitating mass transfer and polyphenol solubilization [31]. In addition, appropriate ultrasonication time enhances acoustic cavitation, leading to cell wall disruption and improved release of intracellular compounds. These results demonstrate that the synergistic combination of DES-8 (ChCl-Fru) and UAE significantly enhances the recovery efficiency of polyphenols from DO.Fig. 33D and 2D response surface curve showing the effects of independent variables on the TPC. Mutual effects of water content in DES and ultrasonic time on TPC (A and D). Mutual effects of liquid–solid ratio and ultrasonic time on TPC (B and E). Mutual effects of liquid–solid ratio and water content in DES on TPC (C and F). 3D and 2D response surface curve showing the effects of independent variables on the TPC. Mutual effects of water content in DES and ultrasonic time on TPC (A and D). Mutual effects of liquid–solid ratio and ultrasonic time on TPC (B and E). Mutual effects of liquid–solid ratio and water content in DES on TPC (C and F). To confirm the validity of the proposed model, verification experiments were carried out under the predicted optimal conditions: ultrasonic time of 39.34 min, water content of 28.6%, and liquid-to-solid ratio of 53.85 mL/g. The experimental TPC was 87.75 ± 0.06 mg GAE/g DO, which was in close agreement with the predicted value of 87.88 mg GAE/g DO, confirming the accuracy and robustness of the model. Extraction kinetics of Dendrobium polyphenols To facilitate the industrial-scale application of DES-UAE for TPC extraction from DO, extraction kinetics experiments were conducted. As shown in Fig. 4A and B, the TPC increased rapidly during the initial stage (0–40 min) and then gradually approached equilibrium. This rapid increase was attributed to the accelerated diffusion of polyphenols from the plant matrix into the solvent, which was significantly intensified by ultrasonic cavitation. The mechanical effects induced by ultrasound, such as microjetting and cell wall disruption, enhanced solvent penetration and promoted mass transfer of intracellular compounds [32]. However, a slight decrease in TPC was observed after 40 min, which can be attributed to the degradation or structural modification of phenolic compounds under prolonged ultrasonication. Excessive ultrasonic exposure may induce localized high temperature and pressure, leading to bond cleavage, oxidation, or transformation of bioactive molecules [16]. Similar phenomena have been reported [6] in the ultrasonic degradation kinetics of DO polysaccharides, where prolonged ultrasonication resulted in molecular structure disruption and functional changes. These findings suggest that, although ultrasound enhances extraction efficiency, excessive treatment may adversely affect the stability of bioactive components.Fig. 4Nonlinear and linear fitting curves of the first-order (A) and second-order (B) kinetic models for TPC extraction using the DES–UAE method. TPC (C), ABTS+ radical scavenging activity (D), and DPPH· radical scavenging activity (E) of Dendrobium officinale extracts obtained using 70% MeOH, 70% MeOH + UAE, DES, and DES + UAE. Nonlinear and linear fitting curves of the first-order (A) and second-order (B) kinetic models for TPC extraction using the DES–UAE method. TPC (C), ABTS+ radical scavenging activity (D), and DPPH· radical scavenging activity (E) of Dendrobium officinale extracts obtained using 70% MeOH, 70% MeOH + UAE, DES, and DES + UAE. Nonlinear and linear kinetic models were used to model the experimental data to better understand the extraction kinetics. Compared with the first-order model, the second-order kinetic model provided a better description of the experimental data, as reflected by the higher R2values under nonlinear (0.9676 vs. 0.9570) and linear (0.9938 vs. 0.8590) fitting conditions (Fig. 4A and B). These findings suggest that the extraction of polyphenols in the DES-UAE system follows a second-order kinetic model, indicating that the extraction rate is governed by both mass transfer and solute–solvent interactions. Notably, the enhanced extraction performance of the DES-UAE system can be attributed to the combined effects of ultrasonic cavitation and the unique solvation properties of DES, which together facilitate cell disruption, improve solvent accessibility, and strengthen intermolecular interactions with target compounds, thereby leading to more efficient and rapid extraction. To quantitatively evaluate the contribution of ultrasound to the extraction process, the extraction efficiencies of DES and 70% methanol systems were compared under both ultrasound-assisted and silent conditions. As shown in Fig. 4C, the application of ultrasound significantly enhanced the TPC and antioxidant activities for both solvent systems. Notably, under silent conditions, DES alone achieved a TPC of 38.37 ± 4.37 mg GAE/g DO, which was already higher than that of 70% methanol under silent conditions (9.32 ± 2.25 mg GAE/g DO), demonstrating the superior solvation capability of DES. Upon ultrasound application, the TPC increased to 88.11 ± 3.09 mg GAE/g DO for DES-UAE, representing a 130% increase compared with silent DES extraction. This enhancement is substantially greater than that observed for the 70% methanol system (107% increase), indicating a pronounced synergistic effect between DES and ultrasound. In addition, as shown in Fig. 4D and E, the radical scavenging activity of phenolic compounds extracted with 70% methanol was lower than that of the DES-UAE extracts. The DPPH and ABTS scavenging rates of the DES-UAE extracts were 90.79% and 96.46%, respectively, both higher than those of the 70% methanol extract (DPPH: 63.94%; ABTS: 69.08%). Under non-ultrasonic conditions, the DES extract exhibited scavenging rates of 70.52% (DPPH) and 82.92% (ABTS), confirming that ultrasound treatment significantly promoted the release of phenolic compounds with antioxidant activity. These results further indicate that the combination of DES and ultrasound exerts a synergistic effect, facilitating the extraction of more phenolic compounds capable of donating electrons to terminate free radical chain reactions. The enhanced performance of DES-UAE can be attributed to the unique physicochemical properties of DES in combination with acoustic cavitation. DESs typically exhibit high viscosity, which limits mass transfer under silent conditions. Ultrasound effectively reduces this limitation through cavitation-induced microstreaming and shear forces, which decrease the apparent viscosity and enhance solvent penetration into the plant matrix [33]. Moreover, the multiple hydrogen-bonding sites in DES facilitate strong interactions with phenolic compounds; ultrasound further promotes these interactions by increasing the frequency of molecular collisions and disrupting boundary layers at the solid–liquid interface [34]. The mechanical effects of ultrasound, including microjetting and cell wall disruption, are particularly effective in the DES system, where the high density and viscosity of the solvent amplify the transmission of acoustic energy [35]. Collectively, these synergistic effects account for the superior extraction efficiency observed for DES-UAE. Molecular dynamics simulation analysis To elucidate the molecular mechanism underlying the enhanced extraction performance of DES system, molecular dynamics (MD) simulations were conducted to investigate the interactions between DES components and three representative phenolic compounds (2-coumaric acid, tristin, and quercetin). Key parameters, including miscibility, solvent-accessible surface area (SASA), hydrogen-bonding characteristics (number, occupancy, and bond length), and non-covalent interaction energies, were systematically analyzed (Fig. 5).Fig. 5Molecular dynamics simulation analysis of interactions between three representative phenolic compounds (2-coumaric acid, tristin, and quercetin) and solvent systems (ChCl–Fru DES vs. 70% methanol). (A) Snapshots of 3 phenolic compounds dissolved in two e solvent systems at 0–100 ns, respectively. (B) Average Solvent-accessible surface area (SASA) of 3 phenolic compounds per system. (C–D) Hydrogen bond number of 3 phenolic compounds in 2 solvent systems at 0–100 ns and corresponding average hydrogen bond number from 20 ns to 100 ns. (E) Hydrogen bond occupancy of 3 phenolic compounds in 2 solvent systems at 0–100 ns. (F) Distance profile of 3 phenolic compounds in 2 solvent systems at 0–100 ns. (G) Average intermolecular interaction energy of 3 phenolic compounds in 2 solvent systems at 20–100 ns. Molecular dynamics simulation analysis of interactions between three representative phenolic compounds (2-coumaric acid, tristin, and quercetin) and solvent systems (ChCl–Fru DES vs. 70% methanol). (A) Snapshots of 3 phenolic compounds dissolved in two e solvent systems at 0–100 ns, respectively. (B) Average Solvent-accessible surface area (SASA) of 3 phenolic compounds per system. (C–D) Hydrogen bond number of 3 phenolic compounds in 2 solvent systems at 0–100 ns and corresponding average hydrogen bond number from 20 ns to 100 ns. (E) Hydrogen bond occupancy of 3 phenolic compounds in 2 solvent systems at 0–100 ns. (F) Distance profile of 3 phenolic compounds in 2 solvent systems at 0–100 ns. (G) Average intermolecular interaction energy of 3 phenolic compounds in 2 solvent systems at 20–100 ns. The miscibility analysis revealed that, during the 0–100 ns simulation, phenolic molecules in the 70% methanol system exhibited high mobility and gradually aggregated into clusters. In contrast, in the ChCl-Fru system, molecular mobility was significantly restricted, and all compounds remained uniformly dispersed and stably surrounded by solvent components (Fig. 5A). This indicates that the DES system effectively suppresses molecular aggregation, thereby enhancing solubility and dispersion stability. The total SASA values of the two systems were comparable (62.18 vs. 61.97 nm2). However, due to the substantial difference in the number of non-aqueous solvent molecules (1650 for ChCl-Fru vs. 3600 for methanol), normalization revealed that the SASA contribution per solvent molecule in the DES system (0.41 nm2) was approximately 3.2 times higher than that in the methanol system (0.13 nm2) (Fig. 5B). This result indicates a stronger effective affinity between DES components and phenolic molecules. Hydrogen-bonding analysis further demonstrated the superiority of the DES system. Within the 20–100 ns interval, the average number of hydrogen bonds in the ChCl-Fru system (33.85) was significantly higher than that in the methanol system (27.58) (Fig. 5C-D). Moreover, the hydrogen bond occupancy in the DES system (17.6%) was markedly higher than that in methanol (0.34%), accompanied by shorter bond lengths (0.2775 vs. 0.2825 nm) (Fig. 5E-F), indicating the formation of stronger and more stable hydrogen-bonding networks. Such dense hydrogen-bonding structures contribute to improved stabilization and solvation of phenolic molecules. From an energetic perspective, all binding free energies were negative, indicating spontaneous interactions. Notably, the average binding energy in the ChCl-Fru system (−2975.55 kJ/mol) was significantly lower than that in the methanol system (−2025.13 kJ/mol) (Fig. 5G), suggesting stronger non-covalent interactions in the DES system. These favorable interactions further promote the homogeneous dispersion and stability of phenolic compounds. Combined with the three-dimensional interaction visualization (Fig. S1), it can be clearly observed that abundant and multi-site hydrogen-bonding interactions are formed between phenolic molecules and DES components, leading to the construction of a dense and stable solvation network. In contrast, fewer hydrogen bonds and weaker intermolecular interactions are present in the methanol system, resulting in a relatively loose solvation structure. These observations provide direct structural evidence that the DES system offers a more favorable microenvironment for phenolic compounds. Overall, the enhanced extraction performance of the DES system can be attributed to synergistic effects, including suppressed molecular aggregation, enhanced solvation capability, strengthened hydrogen-bonding networks, and more favorable non-covalent interaction energies, thereby facilitating the efficient dissolution and stabilization of phenolic compounds at the molecular level. Chemical analysis of Dendrobium phenolic compounds Polyphenolic compounds from DO were extracted using a DESs-UAE and 70% EtOH method and analyzed by UHPLC-Q Exactive Orbitrap HRMS. The base peak chromatogram (BPC) of the extract is shown in Fig. S2, with the extraction results summarized in Table 3. Fourteen flavonoids and ten phenolic acids were identified as the major components, along with four biphenyls, one phenolic aldehyde, one glycoside, and one chalcone, bringing the total to 31 characterized compounds (Table 3 and Fig. S3). Primary identification was based on retention time, precursor ions, and MS/MS fragmentation patterns. These assignments were further verified by comparison with authentic standards, searches in the Compound Discoverer database, and alignment with published literature [36].Table 3The identified chemical constituents from Dendrobium officinale.NoNameFormulaRT (min)Determined (m/z)Error (ppm)MS21SchaftosideC26H28O141.02563.14121.07443.0983,473.1081,383.07722Protocatechuic AcidC7H6O42.99153.01851.96109.02953Salicylic acidC7H6O34.26137.02351.46137.0234,93.03334luteolin-7-O-rutinosideC27H30O164.37609.14670.97285.0410,447.093252-Coumaric acidC9H8O34.40163.03931.84119.049164-Hydroxy-2-methoxybenzoic acidC8H8O44.54167.03432.39152.0133,108.022073-CarboxyphenolC7H6O34.70137.02351.46137.0233,93.03348Syringic AcidC9H10O54.73197.04524.06123.0077,138.03129Rutin*C27H30O164.76609.14681.15463.0882,283.0248,301.0354104-HydroxybenzaldehydeC7H6O24.78121.0294−0.83121.0282,108.020111Vanillic acidC8H8O44.83167.03432.39152.0134,108.022012BiorobinC27H30O154.87593.15193.03413.0878,285.0404134-Coumaric acidC9H8O34.91163.03931.84119.049114ViolanthinC27H30O144.95577.15662.43353.0665,383.0765,413.085415Salicifolioside AC29H38O154.97625.21451.28417.1556,181.0500,387.1089165-Hydroxy-2-methoxybenzoic acidC8H8O45.00167.03421.80108.0219,152.013217VitexinC21H20O105.02431.09873.25283.0600,117.034118DaidzeinC15H10O45.39253.05091.19224.0486,133.0296194-CarboxyphenolC7H6O35.46137.02351.4693.0332,137.023420Quercetin*C15H10O75.48301.03560.66255.0295,273.0411,199.0404,151.050121TristinC15H16O45.59259.09770.77138.0311,170.0213,137.0233,244.074522ApigeninC15H10O55.72269.04581.19225.0553,117.0334,151.002623Naringenin*C15H12O55.78271.06051.47271.0614,151.0027,119.0490,177.018224PadmatinC16H14O75.78317.06683.78193.0135,149.023325Kaempferol*C15H10O65.81285.04081.40285.0405,133.0283,151.002426CerarvensinC20H18O95.91401.08790.25121.0283,313.0720,225.0557,357.061327Okanin 4′-(6′'-acetylglucoside)C23H24O125.93491.11991.02399.6402,146.3369,62.992628Batatasin IIIC15H16O36.12245.11761.63105.0422,151.0754,121.061229GigantolC16H18O46.22275.12780.36137.0581,151.073130ChrysoeriolC16H12O67.08299.05641.00284.0326,240.0417,283.023431Densiflorol AC16H16O47.49271.09791.44228.0401,256.0732,241.0507 Table 3The identified chemical constituents from Dendrobium officinale.NoNameFormulaRT (min)Determined (m/z)Error (ppm)MS21SchaftosideC26H28O141.02563.14121.07443.0983,473.1081,383.07722Protocatechuic AcidC7H6O42.99153.01851.96109.02953Salicylic acidC7H6O34.26137.02351.46137.0234,93.03334luteolin-7-O-rutinosideC27H30O164.37609.14670.97285.0410,447.093252-Coumaric acidC9H8O34.40163.03931.84119.049164-Hydroxy-2-methoxybenzoic acidC8H8O44.54167.03432.39152.0133,108.022073-CarboxyphenolC7H6O34.70137.02351.46137.0233,93.03348Syringic AcidC9H10O54.73197.04524.06123.0077,138.03129Rutin*C27H30O164.76609.14681.15463.0882,283.0248,301.0354104-HydroxybenzaldehydeC7H6O24.78121.0294−0.83121.0282,108.020111Vanillic acidC8H8O44.83167.03432.39152.0134,108.022012BiorobinC27H30O154.87593.15193.03413.0878,285.0404134-Coumaric acidC9H8O34.91163.03931.84119.049114ViolanthinC27H30O144.95577.15662.43353.0665,383.0765,413.085415Salicifolioside AC29H38O154.97625.21451.28417.1556,181.0500,387.1089165-Hydroxy-2-methoxybenzoic acidC8H8O45.00167.03421.80108.0219,152.013217VitexinC21H20O105.02431.09873.25283.0600,117.034118DaidzeinC15H10O45.39253.05091.19224.0486,133.0296194-CarboxyphenolC7H6O35.46137.02351.4693.0332,137.023420Quercetin*C15H10O75.48301.03560.66255.0295,273.0411,199.0404,151.050121TristinC15H16O45.59259.09770.77138.0311,170.0213,137.0233,244.074522ApigeninC15H10O55.72269.04581.19225.0553,117.0334,151.002623Naringenin*C15H12O55.78271.06051.47271.0614,151.0027,119.0490,177.018224PadmatinC16H14O75.78317.06683.78193.0135,149.023325Kaempferol*C15H10O65.81285.04081.40285.0405,133.0283,151.002426CerarvensinC20H18O95.91401.08790.25121.0283,313.0720,225.0557,357.061327Okanin 4′-(6′'-acetylglucoside)C23H24O125.93491.11991.02399.6402,146.3369,62.992628Batatasin IIIC15H16O36.12245.11761.63105.0422,151.0754,121.061229GigantolC16H18O46.22275.12780.36137.0581,151.073130ChrysoeriolC16H12O67.08299.05641.00284.0326,240.0417,283.023431Densiflorol AC16H16O47.49271.09791.44228.0401,256.0732,241.0507 The identified chemical constituents from Dendrobium officinale. Further comparison with literature and databases led to the preliminary identification of five flavonoids as rutin, biorobin, quercetin, padmatin, and kaempferol. Previous studies also indicated that schaftoside, violanthin, vitexin, apigenin, and cerarvensin are derivatives of apigenin and are classified as flavonoids. Additionally, daidzein, naringenin, and chrysoeriol were identified as isoflavones, dihydroflavones, and methoxyflavones, respectively. The phenolic acids were primarily composed of hydroxybenzoic acid and hydroxycinnamic acid derivatives. A total of 10 phenolic acids were identified based on literature and database comparisons. The hydroxycinnamic acid derivatives include 2-coumaric acid (Compound 5: RT 4.40 min, m/z 163.0393) and 4-coumaric acid (Compound 13: RT 4.91 min, m/z 163.0393). The remaining 8 phenolic acids were hydroxybenzoic acid derivatives, namely protocatechuic acid, salicylic acid, 4-hydroxy-2-methoxybenzoic acid, 3-carboxyphenol, syringic acid, vanillic acid, 5-hydroxy-2-methoxybenzoic acid, and 4-carboxyphenol. Moreover, Compounds 21, 28, 29, and 31 were identified as tristin, batatasin III, gigantol, and densiflorol A, all belonging to biphenyls. Three additional compounds were also identified: Compound 10 was 4-hydroxybenzaldehyde, a phenolic aldehyde; Compound 15 was salicifolioside A, a glycoside; and Compound 27 was okanin 4′-(6′'-acetylglucoside), a chalcone. Multivariate statistical analysis and screening of differential metabolites To investigate the differences in phenolic compound content in DO samples from different origins, this study employed a combination of chemometrics and multivariate statistical analysis to systematically analyze 13 batches of DO samples collected from three different regions. PCA score plots were used to effectively demonstrate the metabolic differences between DO samples from various origins. As shown in Fig. 6A and B, the two-dimensional and three-dimensional PCA plots indicate that DO samples from Leqing (LQ) and Hangzhou (LA) in Zhejiang province cluster together, distinctly separated from those originating from Dianan (DN) in Yunnan province. This clustering pattern suggests that samples from the same province tend to group more closely, while samples from different origins show more widespread distribution, with notable clustering differences. These results highlight significant metabolic differences between samples from different regions in terms of metabolite types, quantities, and concentrations. Additionally, sample LQ1 falls outside the 95% confidence interval, likely due to variations in raw materials or storage conditions, marking it as an outlier. Overall, the PCA score plots effectively reveal the metabolic differences between samples from different regions.Fig. 62D and 3D score plot of PCA in different region of Dendrobium officinale (A,B). HCA analysis (C). OPLS-DA score plot between DN with LA (D), DN and LQ (E), DN and LA + LQ. DN, Diannan, Yunnan province; LA, Hangzhou, Zhejiang province; LQ, Leqing, Zhejiang province. 2D and 3D score plot of PCA in different region of Dendrobium officinale (A,B). HCA analysis (C). OPLS-DA score plot between DN with LA (D), DN and LQ (E), DN and LA + LQ. DN, Diannan, Yunnan province; LA, Hangzhou, Zhejiang province; LQ, Leqing, Zhejiang province. The findings were further supported by HCA, which shows that the LQ and LA samples cluster together and are clearly distinct from those of DN (Fig. 6C). To further identify the metabolic differences between DN and other origins, OPLS-DA was conducted based on the PCA results. As shown in Fig. 6D–F, the OPLS-DA model reveals significant separation between DN and the LA, LQ, and LA + LQ groups, confirming notable metabolic differences between these groups. To identify the key differential metabolites, the variable importance in projection (VIP) was calculated, with a threshold of VIP > 1 and p-value < 0.05. The VIP value quantifies the contribution of each metabolite to the classification differences between groups, with higher VIP values indicating a stronger ability to differentiate species and a greater likelihood of serving as potential biomarkers [37]. Using these criteria, 20 differential metabolites were identified between DN and LA, 11 between DN and LQ, and 16 between DN and the LA + LQ combination (Fig. 7A–C). Venn diagram analysis and heatmap visualization revealed 11 shared differential metabolites (Fig. 7D and E). Further structural analysis identified these 11 metabolites as six flavonoids, three phenolic acids, one biphenyl, and one glycoside. These shared differential metabolites play a central role in distinguishing DN from other origins and may serve as potential chemical biomarkers. Notably, similar patterns of differential phenolic metabolites have also been reported in other plant-based systems undergoing processing or fermentation, where dynamic transformations of phenolic compounds significantly contribute to chemical diversity and functional properties. Untargeted metabolomics studies have revealed that such phenolic compounds are highly sensitive to environmental and processing conditions, leading to continuous interconversion, degradation, or biosynthetic shifts [38]. Although the present study focuses on geographical variation rather than processing, these findings suggest that the identified biomarkers may also reflect underlying dynamic metabolic regulation and environmental adaptation mechanisms in DO.Fig. 7The volcano plot of DN and LA (A), DN and LQ (B), DN and LA + LQ (C). Venn diagram of DN vs LA, DN vs LQ, and DN vs LA + LQ (D). Heatmap of different region of Dendrobium officinale (E). The volcano plot of DN and LA (A), DN and LQ (B), DN and LA + LQ (C). Venn diagram of DN vs LA, DN vs LQ, and DN vs LA + LQ (D). Heatmap of different region of Dendrobium officinale (E). Determination of the antioxidant capacity DPPH and ABTS rapid assay methods were employed to evaluate the antioxidant activity of the phenolic extracts. As shown in Fig. 8A and B, phenolic extracts obtained by both extraction methods exhibited concentration-dependent free radical scavenging activity. The free radical scavenging ability of phenolic compounds extracted with 70% ethanol was lower than that of the DES-UAE-treated extracts. Further determination of the IC50 values revealed that the IC50 for the DES-UAE-treated extracts were 65.20 μg/mL for DPPH and 41.80 μg/mL for ABTS, both lower than the 70% ethanol-treated extracts (DPPH: 98.51 μg/mL; ABTS: 78.38 μg/mL). These results indicate that the combination of DES and ultrasound treatment has a synergistic effect, promoting the extraction of more phenolic compounds, which contribute more electrons to terminate the free radical chain reaction. To further evaluate the antioxidant capacity, 13 batches of DO samples extracted using DES-UAE treatment were tested with the DPPH and ABTS assays. The results showed that samples from Leqing and Hangzhou in Zhejiang exhibited weaker antioxidant activity compared to those from Dianan in Yunnan, which demonstrated relatively stronger antioxidant activity (Fig. 8C and D). This difference can be attributed to the higher content of phenolic compounds in samples from southern Yunnan, including 3-carboxyphenol, kaempferol, apigenin, and 4-coumaric acid.Fig. 8Antioxidant activity of Dendrobium officinale polyphenols. DPPH·(A) and ABTS+ (B) radical scavenging rate of DES-UAE and 70% EtOH. DPPH·(C) and ABTS+ (D) radical scavenging rate of different region of Dendrobium officinale. Antioxidant activity of Dendrobium officinale polyphenols. DPPH·(A) and ABTS+ (B) radical scavenging rate of DES-UAE and 70% EtOH. DPPH·(C) and ABTS+ (D) radical scavenging rate of different region of Dendrobium officinale. Conclusion In this study, a novel green and efficient extraction method utilizing DES in conjunction with UAE was successfully developed for the recovery of polyphenols from DO. Among the 24 DES systems evaluated, choline chloride-fructose (ChCl-Fru, 1:2) exhibited the best extraction performance due to its strong hydrogen-bonding network, as confirmed by FT-IR analysis. A systematic optimization strategy combining single-factor experiments, PB design, and BBD was employed to determine the optimal extraction conditions. Under the optimized parameters (ultrasonic time 39.34 min, water content 28.6%, and liquid-to-solid ratio 53.85 mL/g), the maximum TPC reached 87.75 ± 0.06 mg GAE/g DO, with good agreement between experimental and predicted values. Extraction kinetics analysis indicated that the extraction process followed a second-order kinetic model, suggesting that both mass transfer and solvent–solute interactions governed the extraction behavior. MDS further revealed that strong hydrogen-bonding interactions and enhanced solvation capacity of the DES system contributed significantly to the improved extraction efficiency, providing a molecular-level explanation for the DES-UAE mechanism. Chemical profiling using UHPLC-Q Exactive Orbitrap HRMS identified a total of 31 phenolic compounds, including flavonoids, phenolic acids, biphenyls, and other phenolic derivatives. Multivariate statistical analysis demonstrated significant differences in metabolite composition among DO samples from different geographical origins, and 11 key differential metabolites were identified as potential chemical markers. Furthermore, the DES-UAE extracts exhibited superior antioxidant activity compared to conventional ethanol extracts, with lower IC50 values in both DPPH and ABTS assays. Samples from Yunnan showed stronger antioxidant capacity, which was associated with higher levels of key phenolic compounds. Overall, the synergistic effect of ultrasonic cavitation and DES solvation significantly enhanced the extraction efficiency and bioactivity of polyphenols. This study provides a green, efficient, and scalable strategy for the extraction and characterization of bioactive compounds, and offers new insights into the molecular mechanism and functional properties of DES-UAE systems. Although the single-frequency ultrasonic bath employed in this study successfully synergized with DES for polyphenol extraction, recent studies [39] have demonstrated that dual-frequency ultrasonic-assisted extraction exhibits superior efficiency in disrupting plant cell walls and enhancing mass transfer. Dual-frequency ultrasound can intensify cavitation effects through frequency superposition and lower the inertial cavitation threshold, achieving higher extraction yields with lower energy input. Therefore, future work will explore dual-frequency ultrasound to further enhance polyphenol extraction from Dendrobium officinale and facilitate industrial scale-up. CRediT authorship contribution statement Dong-min Cao: Writing – original draft, Visualization, Software, Methodology, Formal analysis, Data curation, Conceptualization. Qian Zeng: Visualization, Validation, Software, Methodology, Formal analysis. Wen-xing Ao: Validation, Software, Formal analysis, Data curation. Shuai Zheng: Visualization, Validation, Software. Ying-xin Lu: Visualization, Software. Xi-cheng Wang: Writing – review & editing, Software, Funding acquisition, Conceptualization. Ming-Hua Xian: Writing – review & editing, Supervision, Resources, Project administration, Conceptualization. Shu-mei Wang: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition, Conceptualization. Funding This research is supported by Projects of the 10.13039/501100001809National Natural Science Foundation of China (grant numbers 82474083, 82173972, 82374172), 10.13039/501100021171Guangdong Basic and Applied Basic Research Foundation (grant numbers 2024A1515012779, 2023A1515011147). Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.