Selective polyphenol extraction during ultrasound-assisted low-temperature extraction of black, white, and green tea

• Multi-response optimisation balanced polyphenol extraction and tannin control • Black tea tolerated high ultrasound intensity when extraction time was constrained • Green tea required intermediate ultrasound intensity to maintain selectivity • White tea exhibited broad process tolerance with limited tannin co-extraction • Contour visualisation show robust operating windows for low-temperature processing Ultrasound-assisted low-temperature extraction is a promising strategy for enhancing the extraction of polyphenolic compounds from tea while maintaining mild processing conditions. However, most existing studies optimise extraction intensity using single-response metrics, even though increased processing severity can promote tannin co-extraction that may compromise sensory quality. This study applied response surface modelling and multi-response optimisation to identify extraction conditions that maximise total polyphenol recovery while minimising tannin extraction during ultrasound-assisted low-temperature extraction of black, white, and green teas. Extraction time (10-30 min), tea-to-water ratio (1.33-5.00 g/100 mL), and ultrasound amplitude (20-70%) were evaluated using a quadratic experimental design. Total polyphenolic and total condensed tannin content were quantified and expressed on a mass-normalised basis (mg GAE/g tea). Multi-response desirability functions were employed to identify robust processing windows rather than single-point optima for each tea. Distinct, tea-specific extraction behaviours were observed. Black tea tolerated higher ultrasound intensities when extraction time was constrained, whereas polyphenol recovery in white tea was favoured by longer extraction times at moderate ultrasound amplitudes. Green tea exhibited the strongest trade-off between polyphenol recovery and tannin co-extraction, requiring intermediate ultrasound intensity and extraction time to maintain selectivity. Response contour plots and sweet-spot visualisation revealed substantial differences in process robustness and sensitivity among tea types, reflecting differences between polyphenol composition and processing histories. These findings show that ultrasound-assisted low-temperature extraction requires specific optimisation strategies that balance extraction efficiency with control of tannin co-extraction. This study provides a framework for designing selective, energy-efficient low-temperature tea extraction processes.