Multivariate analysis of the interactive effects of thermosonication and coagulant type on tofu quality characteristics and yield

Objective This study aimed to model the effects of physical pretreatments of soymilk and different coagulating agents on the physicochemical, textural, yield, microbial quality, and sensory properties of tofu. Methods Soymilk was subjected to three pretreatments: untreated, thermal pasteurization, and thermal high-intensity ultrasound (thermo-ultrasound). Each soymilk treatment was coagulated using magnesium chloride (MgCl 2 ), calcium chloride (CaCl 2 ), or acetic acid (AA). The resulting tofu types were evaluated based on eight physicochemical and quality variables, together with sensory acceptance using a 9-point hedonic scale. Data were analyzed using principal component analysis (PCA), partial least squares (PLS) regression, and multiple linear regression. Results Thermo-ultrasound pretreatment improved tofu hardness, particularly externally, although the effect depended on the coagulant used. Coagulant type significantly affected tofu yield, protein content, and pH. CaCl 2 produced the highest yield (14.6%), followed by MgCl 2 (12.8%) and AA (11.8%), while protein content followed the same trend (10.2%, 8.5%, and 6.6%, respectively). Other compositional parameters remained relatively constant among coagulants. Tofu pH ranged from 6.4 to 7.1 with divalent salts and decreased to 4.7 with AA. Aerobic mesophilic counts ranged from 2.1 to 2.9 log CFU/g, whereas yeast and mold counts remained <2 log(CFU/g). Sensory analysis identified acidity as the most influential factor affecting acceptability (49.38%), followed by consistency and flavor. Thermo-ultrasound reduced soft-texture defects and off-flavors, and ultrasonicated soymilk coagulated with divalent salts achieved the highest acceptance scores (∼6.5, “moderately liked”). Acid-coagulated and untreated samples were least accepted (<3.4). PCA indicated that coagulant type primarily determined macroscopic differences, whereas pretreatment mainly modulated tofu structure. The PLS model with two components (R 2 pred 91%) indicated that hardness is mainly influenced by ultrasound cavitation and pH. Yield depends on moisture content, which is enhanced by pasteurization, with hardness as a key predictor. Achieving optimal tofu balance requires combining ultrasound pretreatment with divalent salt coagulation. Discussion The results show that tofu quality depends on the interaction between soymilk pretreatment and coagulant type. Combining thermo-ultrasound pretreatment with divalent salt coagulation improved textural properties, reduced sensory defects, and produced greater consumer acceptance. Achieving optimal tofu balance requires combining ultrasound pretreatment with divalent salt coagulation.