Aqueous Extraction of Phenolics from Mangosteen Peel Using Ultrasound and High-Shear Mechanical Processing: Extraction Selectivity and Energy Demand

Selecting aqueous extraction technologies for biomass valorization requires balancing process selectivity and energy demand rather than maximizing extraction yield alone. In this study, ultrasound-assisted extraction (UAE) and high-shear mechanical extraction (HSME) were systematically compared for the aqueous recovery of phenolic compounds from mangosteen peel byproduct. Both processes were evaluated in terms of extraction performance, phenolic selectivity, energy demand, temperature rise, and process scalability. While HSME promoted a broader and less selective release of phenolic compounds, reaching total phenolic concentrations of up to 1161 ± 8 μg GAE/mL, UAE favored the enrichment of specific phenolic subclasses, particularly anthocyanin-rich fractions, achieving 97 ± 3 μg C3OG/mL of total anthocyanins and a higher proportion of red pigments (44 ± 1% vs. 38 ± 1% for HSME), and resulting in higher antioxidant capacity despite comparable total phenolic contents. Energy demand was assessed by distinguishing between energy effectively transferred to the medium, determined through a calorimetric approach, and electrical energy required at the system level, revealing contrasting energy delivery modes between shear- and cavitation-driven processes. Although UAE delivered higher specific energy to the extraction medium (up to 130.4 kJ/kg vs. 56.2 kJ/kg for HSME), its batch operation and process complexity impose scalability constraints, whereas HSME exhibited lower energy demand per unit mass and greater operational robustness. A process-level sustainability assessment highlighted clear trade-offs between selectivity, energy demand, and scalability. Overall, the results demonstrate that energy delivery mode is a key design variable in aqueous phenolic extraction, providing a rational basis for selecting extraction strategies according to targeted phenolic profiles and process constraints.