Different methods for natural matter extraction have different extraction yield and efficiency. A comparison of extraction yield of flavonoids between SC-CO₂ extraction and CSE under optimal extracting conditions is shown in Table 4. Table 4 shows that SC-CO₂ extraction could produce an extraction yield of 2.37% ± 0.083%, which was more efficient than CSE (1.51% ± 0.061%). In addition, the extraction time of SC-CO₂ extraction was only 49 min, far lower than CSE (120 min). The results showed that SC-CO₂ extraction was more efficient than CSE. Therefore, SC-CO₂ extraction technique can be recommended as a suitable extraction method to isolate flavonoids from pomelo peel.

The model of scavenging the stable DPPH radicals has been widely accepted as a tool to evaluate the free radical-scavenging activities of materials [22]. Figure 2a describes the scavenging ability of purified flavonoids from pomelo peel on DPPH radicals. At all concentrations tested, flavonoids exhibited a dose-dependent DPPH radical-scavenging activity. The scavenging abilities of flavonoids extracted by SC-CO₂ extraction and CSE on DPPH radicals were from 35.52% to 96.31% and from 21.43% to 73.25%, respectively, with concentration increasing from 12.5 to 200 μg/mL. The EC₅₀ values of scavenging ability on DPPH radicals for flavonoids extracted by SC-CO₂ extraction and CSE were 26 μg/mL and 60 μg/mL, respectively, indicating that flavonoids extracted by SC-CO₂ extraction had a higher scavenging activity on DPPH radicals than that by CSE. The scavenging abilities of the flavonoids extracted by SC-CO₂ extraction and CSE on DPPH radicals were all relatively lower than that of vitamin C (Vc) (EC₅₀ = 22 μg/mL). However, at 200 μg/mL, the DPPH radical-scavenging activity of flavonoids extracted by SC-CO₂ extraction was equivalent to the scavenging activity of Vc used in this study.

ABTS assay is often used in evaluating total antioxidant power of single compounds and complex mixtures of various plants [23]. Figure 2b reveals the scavenging ability of flavonoids from pomelo peel on ABTS radicals. In this assay, the concentration-dependent profile of scavenging activity on ABTS radicals was obvious for the tested flavonoids. The scavenging abilities of flavonoids extracted by SC-CO₂ extraction and CSE on ABTS radicals were from 38.63% to 89.74% and from 25.35% to 75.23%, respectively, with concentration range of 12.5–200 μg/mL. Flavonoids extracted by SC-CO₂ extraction revealed a better antioxidant activity because the EC₅₀ values of scavenging ability on ABTS radicals for flavonoids extracted by SC-CO₂ extraction and CSE were 23 μg/mL and 56 μg/mL, respectively. Although flavonoids of pomelo peel showed a lower scavenging activity on ABTS radicals than Vc (EC₅₀ = 17 μg/mL), the ABTS radical-scavenging activity of flavonoids extracted by SC-CO₂ extraction was to be equal to that of Vc at 200 μg/mL.

Removing hydroxyl radicals is important for the protection of living systems because they are considered to be mainly responsible for the oxidative injury of biomolecules [24]. Scavenging ability of the flavonoids isolated from pomelo peel on hydroxyl radicals is depicted in Figure 2c. In this work, the scavenging abilities of flavonoids extracted by SC-CO₂ extraction and CSE on hydroxyl radicals were concentration-dependent, which were from 21.24% to 68.71% and from 13.72% to 59.33%, respectively, with concentration varying from 12.5 to 200 μg/mL. The EC₅₀ values of scavenging ability on hydroxyl radicals for flavonoids extracted by SC-CO₂ extraction and CSE were 80 μg/mL and 135 μg/mL, respectively, proving that flavonoids extracted by SC-CO₂ extraction were better scavengers for hydroxyl radicals than those by CSE, although the scavenging ability of the flavonoids extracted by SC-CO₂ extraction on hydroxyl radicals was relatively lower than that of Vc (EC₅₀ = 12.5 μg/mL).

The antioxidant capacity of flavonoids from pomelo peel was assessed with the scavenging abilities of DPPH, ABTS and hydroxyl radicals. Our results indicated that flavonoids prepared by SC-CO₂ extraction was found to have a significantly higher antioxidant activity than that obtained by CSE. The antioxidants are believed to donate hydrogen from the phenolic hydroxyl groups and break the free radical chain of oxidation forming a stable end product, which does not initiate or propagate further oxidation [25]. The high antioxidant activity of flavonoids can be attributed to hydroxy groups in the A- and B-rings, and the larger the number of hydroxy groups, the higher is the capacity to scavenge free radicals [26]. For flavonoids extracted by SC-CO₂ extraction, their high antioxidant activity could be attributed to the high presence of phenolic hydroxyl group flavonoids.

Pomelo was purchased from Yuhuan County, Zhejiang Province, China. The peel was washed and dried in an oven at 60 °C for 48 h to a moisture content of 8.0%, then ground to homogeneous powder (0.5 mm diameter) by an electrical food grinder.

Carbon dioxide (99.99% purity) supplied in a cylinder with a dip tube was purchased from Hangzhou Jinggong Specialty Gases Co., Ltd. (Hangzhou, China). ABTS, DPPH and Vitamin C (Vc) were purchased from Sigma Chemical Co. (St. Louis, MO, USA). All other chemicals were analytical grade and purchased from Shanghai Boer Chemical Reagent Co., Ltd. (Shanghai, China).

A supercritical fluid extractor Spe-ed SFE-2 (Applied Separation, USA) was used, which operates with two pumps, a master pump fitted with a cooling jacket on the pump head and a second pump (Knauer pump, model K-501, Berlin, Germany) for the addition of organic modifier. For extraction using SFE, 4 g dried pomelo peel powder was placed in a 30 mL extraction vessel. 85% aqueous ethanol was chosen as a modifier in this study. The static extraction was started when the desired pressure and specified temperature were reached. The operating pressure was provided by an air compressor. The extracted analyte was collected in glass vial with a rubber plug at the top. The CO₂ flow rate was kept at approximately 1.0 mL/min by adjusting the outlet valve manually. The extraction was then performed under various experimental conditions in accordance with the experimental design. In order to assure the accuracy of the experimental data, the experiment was performed in triplicate.

The three independent variables were X₁, X₂ and X₃ representing temperature, pressure and static extraction time, respectively, while the dependent variable was the extraction yield. For statistical calculation, the variables were coded according to:

where x_i is the coded value of an independent variable, X_i is the real value of the independent variable, X₀ is the real value of an independent variable at the center point, and δX is the step change value.

The experimental runs were designed in accordance with a Box-Behnken design with three factors and three levels. According to the single factor experiments, the settings for independent variables were as follows for the low and high values: temperatures of 60 and 80 °C; pressures of 28 and 42 MPa; and static extraction time of 20 and 60 min. Each variable was coded at three levels: −1, 0 and +1. The symbols and levels are shown in Table 5. From the Box-Behnken design, seventeen experimental runs were required, and five replicates at the center (0, 0, 0) of the design were performed to allow the estimation of the pure error sum of squares.

Experimental data were fitted to a quadratic polynomial model using regression coefficients. The generalized quadratic polynomial model used in the response surface analysis was described in Equation 4:

where Y_i is the predicted response, β₀, β_i, β_ii and β_ij are the regression coefficients for intercept, linearity, square and interaction, respectively, while x_i and x_j are the independent code variables. The cross terms represent two-parameter interactions, and the quadratic terms represent second-order non-linearity, while ɛ is the statistical error that represents other sources of variability, such as measurement error.

Ten grams dried pomelo peel powder was placed in a reflux apparatus. Extraction was performed with 150 mL of 85% aqueous ethanol for 120 min at 85 °C. The crude extract was filtered. The solution was concentrated under reduced pressure. The procedure was performed in triplicate.

The content of flavonoids was determined using a method described by Yi et al. [12] with several modifications. Briefly, 1 mL diluted sample was mixed with 1 mL of 5% (w/w) NaNO₂. After 6 min, 1 mL of 10% (w/w) AlCl₃ was added and allowed to stand for 6 min, then 5 mL of 4% (w/w) NaOH was added to the mixture. Absorbance was taken at 510 nm after 15 min. The content of flavonoids was expressed as rutin equivalents through the calibration curve of rutin. The calibration curve (y = 0.5923x − 0.0021, where y is absorbance of sample, x is sample concentration) ranged 10–1000 μg/mL (R² = 0.9996).

The crude flavonoids-enriched extract obtained under the optimized condition was purified using a column (25 × 1.5 cm²) packed with AB-8 macroporous adsorption resin according to the reference [17]. The conditions for purifying the flavonoids by AB-8 resin were: injecting concentration 3.75 mg/mL, pH = 5, injecting velocity 2.0 mL/min, 40% (v/v) ethanol as desorption solvent, desorption velocity of flow 1.5 mL/min. The purified extract of flavonoids was collected and evaporated at 50 °C, and was then freeze-dried for determination of antioxidant property.

In order to evaluate antioxidant activity of flavonoids from pomelo peel, AB-8 macroporous adsorption resin was employed to purify the flavonoids obtained by SC-CO₂ extraction and CSE. The antioxidant activity was investigated using biochemical methods of DPPH, ABTS and hydroxyl radical scavenging assay. Tests were carried out in triplicate.

The DPPH radical-scavenging activity was measured according to the method of Braca et al. [27]. Flavonoid sample with different concentrations (12.5–200 μg/mL, 1.0 mL) was mixed with methanol solution (3.0 mL) containing DPPH radicals (0.2 mM). The mixture was shaken vigorously and incubated for 30 min in darkness at room temperature, and then absorbance at 517 nm was measured.

The scavenging activity of the flavonoids against ABTS radicals (ABTS⁺) was measured using the method of Fellegrini et al. [28] with some modifications. ABTS⁺ were produced by reacting ABTS solution (7 mM, 25 mL) with potassium persulphate (1.4 mM, 0.44 mL), and the mixture was kept in the dark at room temperature for 12~16 h. In the moment of use, the ABTS⁺ solution was diluted with ethanol (475 mL) to an absorbance of 0.70 ± 0.02 at 734 nm. Flavonoid sample with different concentrations (12.5–200 μg/mL, 1.0 mL) was added to ABTS⁺ solution (3.0 mL) and mixed vigorously. After reaction at room temperature for 6 min, the absorbance at 734 nm was measured.

Hydroxyl radical-scavenging activity was determined based on the method described by Smirnoff and Cumbes [29] with some modifications. The reaction mixture contained flavonoid sample with different concentrations (12.5–200 μg/mL, 1 mL) was incubated with a solution containing orthophenanthroline (5 mM, 1 mL), phosphate buffer (7.5 mM, pH 7.4, 0.8 mL) and FeSO₄ (7.5 mM, 0.5 mL). Finally, H₂O₂ (8.8 mM, 0.5 mL) was added, and the reaction mixture was then incubated at 37 °C for 1 h. The absorbance of the resulting solution was measured spectrophotometrically at 532 nm.

The radical scavenging ability was calculated using the following formula: scavenging ability (%) = (1 − A_sample/A_control) × 100, where A_control is the absorbance of control without the flavonoid sample, and A_sample is the absorbance in the presence of the flavonoid sample. The EC₅₀ value (μg/mL) is the effective concentration at which the hydroxyl, DPPH or ABTS radicals are scavenged by 50%. Vc was used as reference compound.

Each experiment was performed three times and the results were expressed as mean ± standard deviation of three replications. The software Design Expert (Trial Version 7.1.3.; Stat-Ease Inc.: Minneapolis, MN, USA) was used for experimental design, data analysis, quadratic model buildings, and graph (three-dimensional response surface and contour) plotting. p value < 0.05 was regarded as significant and p value < 0.001 as highly significant.