The TC berries were harvested in October 2020 in Wuming, China. The TC peels were detached manually and dried in an oven at 37°C for 72 h, and the moisture content of the TC peels was detected with a water activity meter at 4.12 ± 0.16% (METTLER TOLEDO, MJ33, China). The dried peels were placed in a dark at room temperature and then pulverized by the mill through a 60 mesh sieve and stored under a dark desiccator provisional before use. Ultrapure water was prepared using a Milli-Q water purification system (Millipore, Waltham, MA, USA) that was used in the whole study. The TC peel suspension was disrupted by sonication using a Sonics Vibra Cell sonicator (JY92-IIN, Ningbo Scientz Biotechnology Co., Ltd., Ningbo, China) working for 5 s with 10 s intervals and 400 W operation power at 20 kHz for a total of 20 min, and in case of high temperature, ice was placed around the flask aiming to mitigate the destruction of chemical components in TC due to high temperature, the ice was placed around the flask. All chemicals and reagents used in this experiment were of the analytical grades and were purchased from Shanghai Maikelin Biotechnology Co., Ltd, Shanghai, China.

The yield of TCEO was calculated as

The chemical constituents of the TCEO extracted by both UAPE and HD were measured on a 7890A-5975C-mass chromatograph (Agilent, USA) fitted with a split injector. A DB-WAX MS column [60 m × 0.25 mm (i.d.), the film thickness of 0.25 μm] was employed for the separation of the volatile compounds. The programmed column temperature was the initial temperature of 35°C for 5 min, increasing at a constant ramp at 5°C/min to 240°C, then held for 5 min and stopped at an ultimate temperature of 230°C. The gas type was helium, and the flow rate was maintained at a constant of 1 ml/min. The determination process was operated in EI mode with a 70 eV electron energy voltage. The detector was preset at a scan range of 33–500 amu. The injection and interface temperatures were 240 and 80°C, respectively. GC-MS analysis was programmed to perform for a 1.0 μl sample in split mode (split ratio, 1:200). Most components of the EOs extracted by each method were identified on the basis of retention index (Kovat′s RI) and the NIST14 Mass Spectral Library.

The E-nose generally consists of a gas sensor array utilized to detect and distinguish odors of samples through multiple sensors. The discrimination of TCEO extracted by HD and UAPE was measured with an electronic nose. Spiked 500 μl of TCEO was put into a 10 ml flask at 60 °C water bath, which was then removed after 40 min heating and equilibrium for 10 min at room temperature. E-nose conditions were employed as an initial sample flow rate 600 μl/min, sampling interval 1 s, washing time 120 s, detection time 200 s, and trimming zero time 1 s, and each sample was repeated in sextuplicate.

Design Expert 8.0 software (Stat-Ease, Minneapolis, USA) was employed to perform BBD. ANOVA was applied to determine the statistically significant differences between the compared data in the TCEO yield.

All experiments were performed in triplicate. A BBD matrix comprising 29 trials was formulated with Design Expert software. The real responses of each experiment matrix run through BBD are expressed as the average values that were grounded on the built-in default settings in Design Expert 8.0 software, and the others are represented as the mean values ± SD. The level of confidence required for significance was set at p < 0.05. The identified TCEO was analyzed by GC-MS, and the analysis discrimination of TCEO was employed with an E-nose.

The EO extraction is largely linked to the components and materials (24). According to previous studies, plant cell walls were the most resistant to release Eos (25). Therefore, sodium chloride was used to change the cytomembrane permeability (26). The results show that the maximum TCEO yield was obtained at a concentration of sodium chloride 1.5%, followed by the EO yield falling dramatically as the concentration of sodium chloride rose (Figure 2A). Therefore, 1.5% sodium chloride was selected as the optimum concentration for the extraction of the TCEO. The result of the solvent-solid ratio on the yield of TCEO is shown in Figure 2B. The EO yield increased significantly with an increase in solvent-solid ratio of 20:1. Hereby, the 20:1 ratio was selected as the best solvent-solid ratio. Figure 2C shows the effect of extraction time on the yield of TCEO. The TCEO yield increased significantly during extraction time from 60 to 120 min, while, thereafter, no significant difference in the yield of EO was shown. The result indicates that the TCEO has been extracted to the greatest extent at 150 min.

In consideration of the results of the single-factor test, the effects of sodium chloride concentration (X₁), extraction time (X₂), and solvent-solid ratio (X₃) on the TCEO yield were investigated using the BBD (Table 1). A multiple regression fitting to obtain a quadratic polynomial equation that depicts the relationship between each response value and test variables is as follows:

where Y₁ represents the TCEO yield, and X₁, X₂, and X₃ are the actual values of the independent variables. The effects of extraction parameters on the TCEO yield are displayed on the ANOVA table (Table 2); the regression model is significant (p < 0.001), lack of fit is not significant, R² = 99.51%, adj-R² = 98.87%, and the difference between the R² and the adj-R² values observed for all variables indicates that the proposed models are reasonable to fit the experimental data. The t-test and p-values indicate that the independent variable solvent-solid ratio and sodium chloride concentration have a significant effect on the TCEO yield (p < 0.05). The interactive terms with higher F-values and lower p-values exhibit a bulky significance and are capable of verifying a reasonable fitting model. The adequacy precision value of 33.887 indicates that the proposed model could be applied to navigate the design region.

Also, it was observed that the interactive effects of the solvent-solid ratio and sodium chloride concentration (X₁:X₂), solvent-solid ratio and extraction time (X₁:X₃), sodium chloride concentration, and extraction time (X₂:X₃) were significant (p < 0.01). All the three-square terms (X12, X22, and X32) have tremendous significant impacts on the TCEO yield (p < 0.01) and show a parabolic trend in TCEO extraction. Herein, the extraction time (X₃) did not show any statistical significance. Furthermore, a series of three-dimensional (3D) surface plots were generated where the third independent variable was kept constant and varying the other two to determine the interactive effects of the independent variables on the response (27). As shown in Figure 3A, the effect of the solvent-solid ratio showed a quadratic effect on the TCEO yield in a lower solvent-solid ratio range while decreasing with the increasing solvent-solid ratio in a higher range, and the sodium chloride showed a slight quadratic effect on the TCEO yield. A quadratic effect of extraction time and sodium chloride concentration on the TCEO yield was observed as shown in Figure 3B. As depicted in Figure 3C, the solvent-solid ratio exerted a quadratic effect on the TCEO yield in a lower solvent-solid ratio range while decreasing with the increasing solvent-solid ratio in a higher range, and the extraction time showed a slight quadratic effect on the TCEO yield. In summary, the optimal conditions for the traditional HD extraction process were 20.32:1 solvent-solid ratio, 125.38 min extraction time, and sodium chloride concentration 1.15%, yielding a TCEO of 85.69 mg/g. Based on the predicted conditions, the verification experiment was carried out with a 20.32:1 solvent-solid ratio at the sodium chloride concentration 1.15% and 125.38 min extraction time. The TCEO yield was 85.67 mg/g, which was similar to the predicted value, hence conforming to the model validity.

The result of the solvent-solid ratio on the yield of TCEO is shown in Figure 4A. With the increase in solvent-solid ratio, the EO yield increased significantly at the ratio of 20:1. To avoid solution waste and facilitate subsequent operation, 20:1 was selected as the best solvent-solid ratio. The effect of ultrasound time on the yield of TCEO is displayed in Figure 4B; the maximum TCEO yield was obtained at 20 min, followed by the EO yield decrease as ultrasound time extended. The effect of extraction time on the yield of TCEO is shown in Figure 4C. With the increase in extraction time, the TCEO yield increased significantly, and after 60 min, the yield of EO did not change significantly, which indicates that the TCEO has been extracted to the greatest extent. The results show that the maximum TCEO yield was obtained at the concentration of sodium chloride 1.5%, followed by the EO yield falling dramatically as the concentration of sodium chloride rose (Figure 4D). Therefore, 1.5% sodium chloride was selected as the optimum concentration for the extraction of the TCEO.

Ultrasound was used as an assisted method to explore the conditions of TCEO extraction, and the effects of the solvent-solid ratio (X₄), ultrasound time (X₅), extraction time (X₆), and sodium chloride concentration (X₇) on the TCEO yield were observed using a BBD. The results obtained after the 27 runs, along with the complete experimental design, are shown in Table 3. A multiple regression equation was obtained that represents an empirical relationship between the TCEO yield and the process variables as given:

where Y₂ represents the TCEO yield, and X₄, X₅, X₆, and X₇ are the actual values of the independent variables. The effects of extraction parameters on the TCEO yield are displayed on the ANOVA table, as can be seen from Table 4; the regression model is significant (p < 0.001), lack of fit is not significant, R² = 95.44%, adj-R² = 90.87%, and the difference between the R² and the adj-R² values is observed for all variables, indicating that the proposed models are reasonable to fit the experimental data. The t-test and p-values indicate that the independent variable solvent-solid ratio, ultrasound time, and sodium chloride concentration have a significant effect on the TCEO yield (p < 0.05). The interactive terms with higher F-values and lower p-values exhibit a bulky significance and are capable of verifying a reasonable fitting model. The adequacy precision value of 15.187 indicates that the proposed model could be applied to navigate the design region.

Also, it was observed that the interactive effects of the solvent-solid ratio and ultrasound time (X₄:X₅), ultrasound time, and extraction time (X₅:X₆) were significant (p < 0.01). All the four-square terms (X42, X52, X62, and X72) have tremendous significant impacts on the TCEO yield (p < 0.01) and show a parabolic trend in TCEO extraction. Herein, the extraction time (X₆) did not show any statistical significance. Furthermore, a series of 3D surface plots were generated where the third independent variable was kept constant and varying the other two to determine the interactive effects of the independent variables on the response. As shown in Figure 5A, the effect of the solvent-solid ratio and ultrasound time shows a quadratic effect on the TCEO yield in a lower solvent-solid ratio range while decreasing with the increasing solvent-solid ratio and ultrasound time in a higher range. As depicted in Figure 5B, the ultrasound time and extraction time exerted a quadratic effect on the TCEO yield in a lower solvent-solid ratio range while decreasing with the increasing solvent-solid ratio in a higher range. In summary, the optimal conditions for the ultrasound-assisted extraction process were 20.09:1 solvent-solid ratio, 20.7 min ultrasound-assisted, extraction 60.04 min, and sodium chloride concentration 1.62%, yielding a TCEO of 115.072 mg/g. Based on the predicted conditions, the verification experiment was carried out with a 20:1 solvent-solid ratio at the sodium chloride concentration 1.62%, ultrasound-assisted 20.7 min, and extraction 60 min. The TCEO yield was 114.02 mg/g, which was similar to the predicted value, hence conforming to the model validity.

Compared with the extraction yield of traditional HD (85.67 ± 0.73 mg/g), higher EO yields were acquired by the proposed UAPE (114.02 ± 2.79 mg/g) in a shorter extraction time. The extraction time between the two techniques was 120 and 80 min for HD and UAPE, respectively. According to Table 5, a total of 28 constitutions were identified by GC-MS, and the types and percentages of each component of EO are displayed. A comparison of types was mainly terpenes, sesquiterpenes, aldehydes, alcohols, and long-chain terpenes. The percentages of TCEO extracted by the two methods for HD and UAPE were cyclic monoterpenes (90.99, 91.16%), acyclic monoterpenes (5.09, 4.94%), acyclic sesquiterpenes (0.45, 0.47%), aldehydes (1.17, 1.08%), alcohols (0.59, 0.56%), dicyclic monoterpenes (0.64, 0.62%), sesquiterpenes (0.07, 0.02%), cyclic terpene (0.5, 0.66%), and open-chain monoterpene (0.5, 0.53%), respectively. d-Limonene was the predominant constitution in TCEO, occupying 86.38 and 86.75%. The same results were found in a previous study; a variety of citrus were pretreated and grounded, then extracted by HD, supercritical CO₂, molecular distillation, and organic solvent extraction (29–31). Besides limonene, the distribution of other chemical constituents in grapefruit, mandarin, lemon, and sweet orange was found, including the minor components (29).

A higher yield of EO and short time-consuming was obtained by the UAPE technique. A reasonable principle behind such a phenomenon is likely owing to the UAPE being involved in the principle of acoustic cavitation, which promotes the destruction of plant cell walls, favoring solvent penetration, mass transfer, and the release of bioactive compounds quickly from its cells to the collector (32). Concerning traditional HD, the heat kinetic energy diffuses from solvent to the plant materials, resulting in the effective mass transfer occurring from the inner of plant materials to the outside (33). The proposed UAPE displays a significantly higher EO yield than HD, probably owing to the destruction of cellular structures of plant materials (34). However, different EO yields have been reported in previous studies (15). These discrepancies might be due to the differences in geographic location, cultivation, and sample preparation (35). In summary, compared with UAPE and traditional HD, the proposed UAPE was effective at increasing the plant EO yield.

The constituents of the EOs isolated from TC peels by UAPE and HD are presented in Figure 7 and Table 5. The GC-MS analysis results indicated that TCEO comprises various hydrocarbon-based components and metabolites linked with carbonyl and hydroxyl groups. In total, 28 compounds were identified and listed according to their Kovat's index, MS, and relative peak area as given in Table 5. Compared to the EO separated from UAPE and HD, no more obvious differences were found in the components of EOs obtained using the two methods. The presence of these constituents may be due to the long operation time for HD, which results in the materials being fully exposed to water and hence being subjected to oxidization, hydrolysis, and even other reactions. Similar phenomena were observed in the extraction of Cinnamomum camphor EO using a solvent-free microwave-assisted method (36). The significantly higher d-limonene content determined in the EO extracted by HD and UAPE was the most abundant compound in the EO of TC peels with the similarity relative amounts of 86.38 ± 0.31% and 86.75 ± 0.23%, respectively. Notably, the EO extracted using HD and UAPE possessed no significant difference in the high percentage of terpene (97.77% for HD and 97.78% for UAPE).

The type and percentage of terpene detected in the EO for HD and UAPE were cyclic monoterpenes (90.99, 91.16%), acyclic monoterpenes (5.09, 4.94%), acyclic sesquiterpenes (0.45, 0.47%), aldehydes (1.17, 1.08%), alcohols (0.59, 0.56%), dicyclic monoterpenes (0.64, 0.62%), sesquiterpenes (0.07, 0.02%), cyclic terpene (0.5, 0.66%), and open-chain monoterpene (0.5, 0.53%), respectively. A–E represent the chemical materials α-pinene, sabinene, γ-myrcene, d-limonene, and β-phellandrene. E represents the rest of the EO. All the above constituents display no significance (p < 0.05). Several studies regarding the citrus used to extract EO exhibited d-limonene as the predominant component, followed by other components like pinene, linalool, and farnesene (23, 26, 37–39).

The principal component analysis (PCA) plot of the E-nose data is shown in Figure 9. The optimization of the 14 array sensors yielded data with a similar observation in the GC/MS. From the above results, the E-nose proposed in the research was capable of differentiating VOC of TCEO by differential extraction methods. Different extraction methods resulted in volatile aroma compound (VAC) differences. After 14 array sensor optimization, the chart figure displayed in Figure 8 shows that the sensors investigated obvious similarities between the 12 samples. The PCA result shows that the optimization sensors are generally correlated with 59.2% of the total variance containing the first principal component PC1 and the second principal component PC2, represented as PC1 42.5% and PC2 16.7%, respectively.

The PCA score plot shows a clear similarity to HD and UAPE. The VAC differences between the HD and UAPE from the TCEO were mainly reflected in the PC1 axis. Similarly, differences between the extraction methods by HD and UAPE from TCEO were mainly reflected in the PC2 axis. Furthermore, the difference between the 12 samples displayed in the chart graph (Figure 8) was similar. From the above result, we show that the UAPE treatment has less effect on the flavor of TCEO compared with the HD technique. Thus, ultrasound pretreatment was an important factor to be considered when designing EO extraction. Such a phenomenon was found in a previous study (29).

E-nose had potentially been used to preliminarily discriminate the VAC through sensor response analysis (40). Radar chart analysis and PCA are displayed in Figures 8, 9. From Figure 8, all the response sensors can be described as sensitive E-nose to detect VAC, wherein the extraction methods by the proposed UAPE and HD showed no significance. The principal component of the analysis chart is displayed in Figure 9. The TCEO extraction methods of UAPE and HD were densely distributed together and overlapped, illustrating no significant differences, and showed some similarities to the result of GC-MS (Table 5).