Materials and chemicals

Fresh passion fruit peels were collected at a juice shop in Thu Duc city, transferred to the laboratory, washed and dried on the same day at 60°C overnight so that the sample moisture was less than 10%. Afterward, the dried samples were ground and sieved through 500 µm mesh to obtain a uniform powder. The passion fruit peel powder (PFPP) was collected, sealed in small bags (50 g each), and stored in a refrigerator for further use. Chemicals of analytical grade were used for extraction and analyses.

Ultrasound-assisted extraction

The ultrasound-assisted extraction was carried out by adapting the approach of Ahmad et al (2015)13. In detail, various amounts of PFPP were mixed with 20 mL of sodium acetate buffer (pH 5) to achieve different solvent-to-solid ratios ranging from 30-70 mL/g. The mixtures were then treated in an ultrasonic bath (WUC-A10H, South Korea) at a frequency of 40 kHz in the temperature range of 40-70°C for 40-70 min. After treatment, the mixtures were quickly cooled to ambient temperature and centrifuged (Z326K, Germany) for 15 minutes at 4°C and 4000 rpm. The supernatants were collected, mixed with 96% ethanol at a ratio of 1:10 (v/v) and kept overnight in a refrigerator for complete precipitation. The precipitated crude polysaccharides were obtained by filtration, and the filtrates were collected for polyphenol recovery.

Box–Behnken design and regression analysis

A Box‒Behnken factorial design (BBD) was employed for the optimization of UAE with three variables: the solvent-to-solid ratio, ultrasonication temperature and duration. Table 1 presents the symbols, units, and coded and true levels of these three variables. The design included 12 factorial points (1, +1) and 5 central points (0), while the entire set of tests comprised 17 runs, which were conducted in a random order with three replicates.

The obtained data were fitted to a second-order polynomial equation (quadratic model) as described in Eq. (1) to correlate the relationships between the independent variables and the response:

where is the response for either PS or TPC; symbolizes the coefficients; and represents the coded independent variables.

To assess the statistical significance of the developed model, the value, value, coefficient of determination (), adjusted (), and predicted () were used. The information was then used to create a 3-D response surface. The desirability function methodology was utilized to estimate the optimal extraction conditions.

Analytical methods

Statistical analysis

Each experiment was performed in triplicate, and the experimental data are expressed as the mean ± standard deviation. Design-Expert software (Trial version, Stat-Ease Inc., USA) was used for ANOVA and optimization.

Box‒Behnken design and regression analysis

Table 2 presents the experimental data for the BBD matrix with PS yield and TPC as the response. After 17 runs, the PS yield ranged from 8.68 to 36.13%, while the TPC varied between 11.55 and 47.51 mg GAE/g. To investigate the combined effects of independent variables (i.e., the solvent-to-solid ratio, ultrasonic temperature and duration) on the PS yield and TPC from PFPP, quadratic models were constructed with the linear and quadratic terms of each variable and their interactions. Table 3 provides the ANOVA results used to evaluate the models. Both models for PS yield and TPC were highly significant, with p values of < 0.0001 and a nonsignificant lack of fit (p values > 0.05), showing the adequacy of pure error. The coefficients of determination (R) were determined to be > 0.998, indicating that the formulated models could explain more than 99.8% of the variability. Furthermore, the predicted R values > 0.97 were reasonably consistent with the adjusted R value of 0.99. Desirable Adeq. Precision values greater than 4 also indicated appropriate signals for the models. Second-order polynomial models representing the correlation between the three independent variables (in their coded levels) and responses were produced in Equations (2) and (3) as follows:

where Y and Y are the responses (PS yield and TPC, respectively), and X, X and X are the independent variables, i.e., the solvent-to-solid ratio, ultrasonic temperature and duration, respectively.

The predicted data of the responses obtained from the two models are presented in Table 2 for comparison with their experimental values. The results in Table 3 indicate that all three variables had significant effects on both responses at their linear and quadratic levels (X and X), with p values < 0.05. On the other hand, the interaction effects of XX and XX were significant (p < 0.05), and those of XX were negligible (p > 0.05) for both responses.

3D-surface responses

To further understand the interaction of variables, 3D response surface graphs (Figure 1) were generated by plotting the response against two independent variables while holding the third constant at its zero level. The images illustrated that both the PS yield and TPC were low at the lowest solvent-to-solid ratio (30 mL/g). These responses markedly increased with increasing solvent-to-solid ratio but slightly decreased at the highest concentration of 70 mL/g. These observations align well with the principles of mass transfer, which suggest that the concentration gradient between the solid and the solvent drives the transfer of mass15. A higher solvent-to-solid ratio amplifies this gradient, accelerating the diffusion rate of chemicals from the solid material into the solvent. However, it also prolongs the time needed to achieve equilibrium. The solvent-to-solid ratio can profoundly influence the equilibrium constant, revealing a relationship between yield and solvent consumption characterized by an exponential increase followed by a plateau as the maximum yield approaches16.

Similar trends were also observed for the effects of ultrasonic temperature and duration. A lower ultrasonic temperature could reduce the solubility of the target compounds in the solvent, leading to insufficient extraction efficiency17. Furthermore, some plant materials may require higher temperatures to effectively breakdown cell walls for the release of their internal substances. However, at elevated temperatures (higher than 60°C in this study), both responses decreased with increasing temperature. This may be due to membrane denaturation at high temperatures, causing difficulty in substance diffusion into the solvent, or due to the instability of phenolic compounds at high temperatures18. On the other hand, increasing the ultrasonication duration to less than 60 min could improve the extraction yield by softening plant tissues, weakening cell wall integrity, and hydrolyzing phenolic-protein, polysaccharide-protein, and phenolic-polysaccharide complex bonds, as well as increasing the solubility of target compounds in the solvent19. In contrast, extending sonication beyond 60 minutes resulted in a lower extraction efficiency for PS yield and TPC. This could be attributed to structural alterations in polyphenols20 or polymeric breakdown of polysaccharides21.

Optimization and validation

The trade-offs among numerous variables were balanced to simultaneously optimize two responses, i.e., PS yield and TPC. The results in Table 4 present the optimal conditions, including a solvent-to-solid ratio of 53.9 mL/g, an ultrasonication temperature of 57.6°C and an ultrasonication duration of 57 min. Meanwhile, the predicted optimal response values were 36.46% and 48.35 mg GAE/g for the PS yield and TPC, respectively. The data obtained from the experiment under the optimal conditions with minor modifications to the variable levels, as shown in Table 4, aligned well with their predicted values, which were 35.76 ± 1.54% and 47.51 ± 1.77 mg GAE/g for PS yield and TPC, respectively. This could confirm the adequacy and significance of the models. Compared with the efficiency of individual extraction, the yield of polysaccharides in this study was greater than that previously reported by Pereira et al. (2024)22 using pressurized solvent extraction, by Vasco-Correa and Zapata (2017)23 using enzymatic extraction, or by Kulkarni and Vijayanand (2010)2 using the conventional method (28%, 26% and 15%, respectively). Moreover, the TPC in this study seemed to be slightly greater than that recorded by Wang et al. (2021)24 using the cellulase-assisted extraction method or by Vo et al. (2023)25 using UAE under milder conditions (liquid-to-solid ratio of 28 mL/g and 20 min) for single extraction (22.34 mg GAE/g and 39.38 mg GAE/g, respectively). The difference may be due to variations in the employed methods, extraction conditions and materials. For example, in this study, pH 5 buffer was used as the solvent for the extraction. Although water extraction has been applied as the traditional method for the extraction of natural polysaccharides, acidic environments have been demonstrated to enable more effective cleavage of glycoside bonds, resulting in higher yields of bioactive low-molecular-weight polysaccharides26, 27, 28. In addition, although various solvents, such as ethanol, methanol, or natural deep eutectic solvents, are commonly used to extract phenolics, acidic conditions have been revealed for their ability to hydrolyze glycoside bonds in phenolic derivatives and transform them into free phenolics for easier release29, 30, 31. Therefore, the use of acidic buffer may be effective for the coextraction of both polysaccharides and polyphenols. In conclusion, these comparisons implied the potential benefits of simultaneous extraction of polysaccharides and polyphenols using pH 5 buffer with the UAE method.