Accelerating the coagulation of germinated soymilk by glucono-δ-lactone using different physical treatments and quality assessment of separated soybean whey

Abstract

This investigation was undertaken to assess the efficacy of glucono-δ-lactone coupled with thermal (steaming and microwave) and non-thermal (ultrasound) treatment in the coagulation of soy protein from germinated soybean seeds. The findings indicated that both microwave and ultrasonic treatments significantly accelerated coagulation of soy protein (5.4–8.7 times faster) compared to conventional steaming treatment. Regarding textural properties, soybean curd obtained from germinated soymilk displayed increased hardness values (122.0 to 277.3 g) and microwave treatment further augmented this textural attribute, except for the sonicated samples. FTIR data also confirmed that the microwave treatment was found to elevate the β-sheet content while concurrently diminishing the α-helix content within the soybean curd. Concerning the soybean whey isolated from soybean curd, the germination was shown to amplify the concentrations of the amino acids Arg, Glu, and Asp. Although the phenolic content in the whey fraction of the germinated samples was higher than that of the non-germinated samples, the FRAP activity of the microwave- and ultrasound-treated samples followed the opposite trend. Overall, microwave treatment reduced processing time while improving texture of tofu, suggesting that microwave treatment can serve as an effective processing intervention to improve both efficiency and structural properties in tofu manufacturing.

Highlights

• •Microwave and ultrasound accelerated coagulation by 5.4–8.7 times than steaming.
• •Tofu from germinated soybean seeds had higher hardness than non-germinated seeds.
• •Microwave improved the hardness of tofu coagulated with GDL.
• •Microwave increased the β-sheet while decreased the α-helix content of soybean curd.
• •Germination intensified the Arg, Glu and Asp amino acids content in soybean whey.

1. Introduction

Tofu, a traditional food product originating from East Asia, represents a significant protein source in global plant-based diets (Qin et al., 2022). Its production fundamentally relies on the controlled coagulation of soybean proteins, primarily glycinin and β-conglycinin, which constitute approximately 80% of the total protein content in soybeans. The coagulation process involves complex physicochemical interactions between these proteins and various coagulating agents, resulting in the formation of a three-dimensional gel network that entraps water, lipids, and other components (Guan et al., 2021). The mechanism of soybean protein coagulation can be categorized into two primary pathways: salt-induced and acid-induced. Salt coagulants, such as calcium sulfate (CaSO₄) and magnesium chloride (MgCl₂), function through ionic cross-linking of protein molecules, whereas acid coagulants, including glucono-δ-lactone (GDL), induce protein denaturation by reducing the pH below the isoelectric point of soy proteins (approximately pH 4.5) (Shi et al., 2020). GDL represents a significant acidulant with unique hydrolysis kinetics that facilitate controlled protein precipitation. GDL functions through a time-dependent hydrolysis mechanism, gradually releasing gluconic acid that incrementally reduces pH and induces protein denaturation without the abrupt precipitation characteristic of direct acid addition (Li et al., 2022). This controlled acidification promotes the formation of fine-stranded protein networks with distinctive textural properties, yielding a homogeneous gel structure with smooth mouthfeel and enhanced water retention capacity. The resultant tofu exhibits characteristic physicochemical attributes that distinguish it from products coagulated by traditional mineral salts such as calcium sulfate or magnesium chloride (Shi et al., 2020). Process parameters, including coagulant concentration, coagulation temperature, and agitation conditions, critically affect protein-protein and protein-water interactions during gel formation (Geng et al., 2024; Khoder et al., 2024; Li et al., 2022). The optimization of these parameters represents an ongoing challenge in tofu manufacturing, as they directly impact product yield, texture, water-holding capacity, and sensory attributes.

The coagulation of soybean proteins through physical treatments such as microwave and sonication has been extensively studied, revealing distinct effects on protein structure and functionality. Microwave treatment, as explored in several studies, has been shown to enhance the solubility and functional properties of soybean proteins. Similarly, microwave treatment of glycosylated soybean 7S proteins for 60 s resulted in a more ordered protein structure, enhancing water-holding capacity, emulsification, and foaming properties (Zhang et al., 2025). Moreover, microwave processing has been effective in inactivating trypsin inhibitors in soybeans, thereby improving protein digestibility (Vagadia et al., 2018). On the other hand, ultrasonic treatment has been shown to induce structural unfolding and exposure of hydrophobic groups in soy protein isolates, which enhances solubility, emulsifying, and foaming properties (Kong et al., 2023). Ultrasonic treatment also significantly improves the gelling ability and structural properties of SPI, making it suitable for applications like tofu production (Wu et al., 2024). The combination of ultrasound and microwave technologies has been proposed to further enhance food processing efficiency by improving mass transfer and heating effects, suggesting potential for more comprehensive applications in food processing (Zhou et al., 2024). Additionally, physical processing techniques like high-pressure treatments and sonication can reduce soybean allergenicity while maintaining nutritional and sensory properties, which is crucial for consumer health and product quality (Kerezsi et al., 2022). These treatments can also affect the microstructure and digestibility of tofu, as seen with GDL-induced gels that retain bioactives better than other coagulants, thus enhancing the delivery of bioactive compounds during digestion. The combination of GDL with these innovative physical treatments not only optimizes the coagulation process but also tailors the nutritional and sensory attributes of tofu, making it a versatile and health-conscious food product (Geng et al., 2024).

Soybean whey, a by-product of tofu production, offers numerous nutritional and functional benefits that can be harnessed in various applications. It is rich in proteins, simple sugars, oligosaccharides, minerals, and soy isoflavones, making it a valuable substrate for fermentation and other processes (Chua & Liu, 2019). Additionally, soy whey can serve as a prebiotic due to its non-digestible oligosaccharides, which promote the growth of beneficial lactic acid bacteria in the colon, potentially enhancing mineral absorption (Tenorio et al., 2010). The use of fermented soybean whey as a coagulant in tofu production has also been shown to improve the protein and amino acid content, as well as the textural and sensory qualities of tofu, making it a preferred choice among consumers (Shi et al., 2020). Tofu is conventionally manufactured mostly from soybean seeds. Germinated seeds possess potential as a novel raw material for tofu manufacturing, attributable to the benefits conferred by the germination process. The germination process significantly enhances the bioavailability of nutrients in soybean seeds by altering their chemical composition and reducing antinutritional factors. During germination, the isoflavone content in soybeans increases, with the highest bioaccessibility observed in daidzein, genistein, and glycitin, particularly after five days of germination, which also enhances antioxidant capacities (Lu et al., 2024). Germination also leads to a reduction in phytic acid due to the activation of phytase, thereby increasing the availability of minerals such as phosphorus, calcium, magnesium, and potassium (Dong & Saneoka, 2020). Additionally, germination improves the digestibility of proteins by activating proteases and reducing protease inhibitors, which enhances protein breakdown and increases free amino acids, notably asparagine. The process also boosts the levels of vitamins such as ascorbic acid, riboflavin, and niacin, contributing to the nutritional value of the germinated seeds (Idowu et al., 2020). Germination of soybeans enhances their nutritional profile by increasing the bioavailability of nutrients and reducing antinutritional factors, which improves protein digestibility and the production of bioactive peptides with antioxidant and antidiabetic properties (Fadlillah et al., 2025; González-Montoya et al., 2018; Wang et al., 2024). These peptides, derived from proteins like glycinin and β-conglycinin, exhibit activities such as inhibiting dipeptidyl peptidase-IV, α-amylase, and α-glucosidase, which are crucial for managing diabetes (González-Montoya et al., 2018). Additionally, germination increases the content of beneficial compounds like phenolics and flavonoids, enhancing antioxidant capacity and potentially offering protective effects against diseases like Alzheimer's (Shabbir et al., 2022). Furthermore, this process improves the functional properties of soy protein, such as solubility and emulsifying capacity, which are crucial for effective coagulation during tofu production (Aijie et al., 2014; H. Yang et al., 2017). The reduction in trypsin inhibitor activity and the breakdown of certain protein subunits during germination contribute to a more efficient coagulation process, resulting in tofu with improved texture and nutritional quality (Aijie et al., 2014). The use of germinated soybeans in tofu production has been shown to yield tofu with higher protein content and improved sensory attributes, such as taste and flavor, although it may result in a softer texture compared to tofu made from ungerminated soybeans (Murugkar, 2014).

Currently, there is little information regarding the coagulation efficiency of soymilk obtained from germinated soybean seeds using different physical treatments combined with GDL. Therefore, this study was conducted to compare the coagulation efficiency by different physical treatments including steaming, microwave treatment and sonication on the coagulation time, texture and secondary structure of soy proteins from non-germinated and germinated soybean seeds. In addition, the chemical composition and antioxidant capacity of whey separated from soybean curd were also evaluated.

2. Materials and methods

2.1. Materials and chemicals

Soybean seeds utilized in this investigation were HLDDN910 cultivar and purchased at the local market in Ho Chi Minh city (Vietnam) and stored in bags at ambient temperature, away from sunshine and moisture. The original soybeans exhibited the following proximate composition: moisture 7.77%, total lipid 32.56%, crude protein 37.22%, ash 5.99%, and total carbohydrate 16.46%. Other chemicals were of analytical grade.

2.2. Coagulation of soybean curd and whey separation

Upon acquisition, soybeans were sorted to remove contaminants such as dirt and decayed kernels, then sanitized by immersion in a 0.1% KMnO₄ solution for 3–4 min. The seeds were subsequently soaked in 0.4% sucrose for 4 h and rinsed with water prior to germination to enhance the biosynthesis of bioactive compounds during the germination process (Liu, Sun, et al., 2024). Soybeans were germinated in darkness for two days at 25 °C, after which the seed coat and hypocotyl were removed for collecting cotyledon with moisture content of 70.43%. Cotyledons were subsequently steamed for 15 min, rapidly cooled with water, ground with water (1:1 ratio), filtered through cheesecloth, and then adjusted to 17% total solids. Before the coagulation, the soymilk samples were subjected to different treatment, including steam treatment for 10 min (H samples), microwave treatment at 2780 W for 40 s (MW samples), and sonication at 200 W for 15 min (S samples) according to previous studies (Olaniran et al., 2025; Xin et al., 2025). The treated soymilk samples were cooled to 30 °C prior to coagulation with GDL.

The coagulation procedure was executed by adding 100 mL of soy milk to a container holding 3 mL of a 0.15 g/mL GDL solution. Sample coagulated with 3 mL of 0.15 g/mL calcium chloride solution served as the control sample. After coagulation, the soybean curd was kept on a layer of cheesecloth for 1 day at 4 °C to obtain soybean whey and final tofu with firmer texture. Table 1 presents the samples' code in this investigation.

Table 1.

Sample coding and description.

Sample code	Germination time (days)		Physical treatment			Coagulant
	0	2	Heat (steam)	Microwave	Sonication	GDL	CaCl2
Ca_0	✓		✓				✓
Ca_2		✓	✓				✓
H_0	✓		✓			✓
H_2		✓	✓			✓
MW_0	✓			✓		✓
MW_2		✓		✓		✓
S_0	✓				✓	✓
S_2		✓			✓	✓

2.3. Characterization of soybean curd

2.3.1. Coagulation time

The coagulation time (s) was defined as the duration from the addition of GDL until the soy milk became solid curd when tilted at 45°, and the internal gel structure was evaluated by probing the curd with a wooden stick inserted through the surface of the curd.

2.3.2. Texture profiles

The textural properties of soybean curd were measured by CT3 texture analyzer (Brookfield, USA) using TA4/1000 cylindrical probe. The test conditions were probe travel distance of 50 mm, activation force of 5 g and test speed of 2.0 mm/s. The textural parameters including hardness, adhesiveness and springiness were obtained by CT13 Pro software.

2.3.3. Color attributes

The color index of soybean curd samples was determined by using a handheld colorimeter model NR110 (3NH, China). The device was calibrated with a white ceramic piece and the CIELab color system with color quantitative parameters including L*, a*, b* was recorded with three replicates.

2.3.4. SDS-page

SDS-PAGE electrophoresis was conducted to examine the protein content of hydrolysate samples with a 10% SDS-PAGE gel. Before electrophoresis, the soybean curd samples were denatured by combining them with a 3× loading buffer in a 4:1 ratio, incubating at 100 °C for 3 min, chilling for 2 min, and repeating this denaturation procedure three times. A 20 μL aliquot of the denatured protein solution was applied to a 4% stacking gel (comprising 30% acrylamide, Tris-HCl buffer at pH 6.8, water, 10% SDS, 10% APS, and TEMED) and subjected to separation at 80 V. The proteins were further separated in a 10% resolving gel, consisting of 30% acrylamide, Tris-HCl buffer at pH 8.8, water, 10% SDS, 10% APS, and TEMED, with a current of 120 V. Subsequent to electrophoresis, the gel was subjected to staining with 0.23% Coomassie Blue R-250 for 30 min and subsequently destained using a solution of 10% methanol and 10% acetic acid until clarity was achieved. The gel was recorded with the BioDoc-It UVP imaging equipment (Analytik Jena AG, Germany). Protein bands manifested as blue-hued bands and were evaluated against molecular weight standards spanning from 15 to 250 kDa.

2.3.5. Secondary structure of protein by FTIR

Fourier transform infrared (FTIR) spectroscopy was performed to characterize the secondary structure of the protein. FTIR spectra of the samples were collected in the wavelength range of 600–4000 cm⁻¹ using a MIR Frontier NIR spectrometer (Perkin Elmer, UK) at a resolution of 4 cm⁻¹. Deconvolution, peak splitting, and spectral fitting of Amide I bands were performed to obtain the proportions of protein secondary structures including α-helix, β-sheet, β-turn, and random coil. Specifically, the amide I region (1600–1700 cm⁻¹) should be subjected to Fourier self-deconvolution (FSD) followed by second derivative analysis to identify component peak positions. The amide I region (1600–1700 cm⁻¹) was baseline-corrected using a second-order polynomial function, followed by Fourier self-deconvolution and second derivative analysis to identify component peaks. Gaussian curve fitting with fixed peak centers (1625, 1635, 1648, 1654, 1665, and 1682 cm⁻¹) and constrained bandwidth (±2 cm⁻¹) was performed using OriginPro. The relative content of each secondary structure was calculated based on the integrated peak area normalized to the total amide I area. Peak assignments in the amide I region should follow established conventions: α-helix (≈1650–1658 cm⁻¹), β-sheet (≈1620–1640 cm⁻¹; with high-frequency components around ≈1670–1690 cm⁻¹ often associated with antiparallel β-sheets), β-turns (≈1660–1680 cm⁻¹), and random coil (≈1640–1650 cm⁻¹, typically ∼1648 ± 2 cm⁻¹). Fitting constraints should include fixed peak width ranges (typically 10–30 cm⁻¹) and iterative optimization until residual errors are minimized (R² > 0.99). The relative area of each fitted peak should then be calculated as a percentage of total amide I band area, ensuring all components sum to 100%.

2.4. Characterization of soy whey

2.4.1. pH, total soluble solids (°brix), and turbidity

pH of whey samples was measured using a Mi150 pH meter (Milwaukee, Romania). The total soluble solids (°Brix) of whey samples were measured using a Master-53 M handheld refractometer (Atago Ltd., Tokyo, Japan). The turbidity of tofu whey samples was measured using the MI415 portable turbidity meter (Milwaukee, Romania).

2.4.2. Amino acid profiles

The sample's amino acid composition was examined utilizing the L-8900 automated AAA system (Hitachi, Tokyo, Japan), interfaced with an HPLC column containing ion-exchange resin 2622 PF (60 × 4.6 mm, 3 μm). Sample preparation according to the manufacturer's guidelines, included protein hydrolysis, pH correction, and filtering. Cystine and methionine were distinguished from the other 17 amino acids regarding sample processing. These two amino acids were specifically oxidized with performic acid, subsequently hydrolyzed with HCl for 24 h, adjusted to pH 2.2 using citrate buffer, and filtered through a 0.45 μm membrane. A 10 μL filtered sample was fed into the system at a column temperature of 57 °C and analyzed with five mobile phases utilizing ninhydrin reagent as per the manufacturer's guidelines.

2.4.3. Total phenolic content

The phenolic content was evaluated utilizing the Folin–Ciocalteu technique as described by (Nguyen et al., 2022). Briefly, 0.6 mL of a whey sample was diluted with distilled water and combined with 1.5 mL of Folin-Ciocalteu reagent. After incubation for 5 min, the mixture was combined with 1.2 mL of 7.5% Na₂CO₃ solution and allowed to react for 30 min. The absorbance of the mixture was recorded at 765 nm utilizing the UV-1800 spectrophotometer (Shimadzu, Japan) against water as blank. The total phenolic content was quantified as mg of gallic acid equivalent (mg GAE/L).

2.4.4. FRAP activity

FRAP analysis was performed according to the method of (Nguyen et al., 2022) with some modifications. The FRAP reagent was prepared as follows: 100 mL of 0.3 M acetate buffer solution at pH 3.6, 1 mL of 0.01 M TPTZ prepared in 0.04 M HCl, and 1 mL of 0.02 M FeCl₃. For the reaction, 0.2 mL of whey sample was mixed with 2.8 mL of FRAP reagent in a test tube; the mixture was then left to react for 30 min in the darkness. The absorbance was measured at 593 nm against water as blank and FRAP activity was expressed as mg Trolox equivalent per liter (mg TE/L).

2.5. Statistical analysis

All statistical methods, including the normality test (Shapiro-Wilk test), homoscedasticity of variances (Levene's test), one-way ANOVA, and post-hoc Tukey test, were conducted at a 5% significance level utilizing R version 4.4.2 (stats, car, agricolae packages). Pearson's correlation analysis was performed in R using the base stats package, specifically the cor() and cor.test() functions. All charts were generated using OriginPro 2025 (OriginLab, USA) and the data were presented as mean ± standard deviation of five replicates (n = 5).

3. Results and discussion

3.1. Effects of treatment methods on coagulation performance and soybean curd properties

3.1.1. Coagulation time

Fig. 1 illustrates the impact of various physical treatments on the coagulation time of soy protein using GDL as a coagulant, in contrast to calcium salt as a control. The findings indicated that microwave or ultrasonic treatment markedly enhanced the coagulation time of GDL, with values of 151.6–164.2 s and 121.5–144.3 s, respectively, in contrast to the sample subjected to conventional steam treatment (891.0–1055.2 s). Apparently, the sample using calcium salt as coagulant had the fastest coagulation time (11.4–11.6 s). Microwave energy directly heats water molecules within the soymilk, resulting in faster and more uniform temperature elevation compared to conventional heating methods. This rapid temperature increase accelerates the denaturation kinetics of soybean proteins (Varghese & Pare, 2019). Microwave radiation can also induce non-thermal effects on protein structure, potentially altering the exposure of hydrophobic domains and facilitating protein-protein interactions during coagulation (Wang et al., 2023). When used in conjunction with GDL or other acidulants, microwave treatment can facilitate hydrolysis reactions and acid production, expediting pH reduction and subsequent protein gelation. For sonication treatment, ultrasound promotes partial unfolding of native protein structures, exposing reactive groups that participate in cross-linking during gelation (Mozafarpour & Koocheki, 2023). Ultrasonic treatment reduces the size of colloidal particles in soymilk, creating more reactive surface area for coagulant interaction and potentially yielding finer gel networks (Gao et al., 2021). It can be concluded that sonication improves the diffusion of coagulants throughout the soymilk system and potentially reduces coagulation time. The rapid coagulation achieved through microwave treatment satisfies industrial throughput requirements, representing a six-fold reduction compared to conventional steaming, thereby offering substantial improvements in production efficiency, energy consumption, and processing capacity for commercial tofu manufacturing operations.

Fig. 1.

Coagulation time of soybean curd of different germination time and treatment methods. Sample abbreviation: X_a; where X is the treatment methods (Ca, H, MW, and S are calcium salt, heat, microwave and sonication treatment) and a is the germination time (0 and 2 days).

It is noteworthy that soymilk derived from germinated seeds has a longer coagulation time than the original sample. However, Liu, Zhou, et al. (2024) indicated that germinated soybean milk typically exhibits faster coagulation compared to non-germinated counterparts when subjected to the same coagulant concentration and conditions. This can be attributed to the difference in soybean variety and treatment conditions among various studies. Germination processes induce conformational changes in protein secondary and tertiary structures, potentially exposing previously buried hydrophobic domains and altering surface charge distribution (Aijie et al., 2014).

While GDL was added in equal amounts across all treatments to maintain experimental consistency, the lack of time-resolved pH data limits a comprehensive understanding of the acidification kinetics and its direct relationship with gelation behavior. Since GDL hydrolysis rate can be influenced by temperature and the prior physical treatment applied to soymilk (Li et al., 2024), differences in pH evolution among treatments may have contributed to variations in coagulation outcomes that were not fully captured in this study. Incorporating continuous pH measurements in future work would enable a more mechanistic interpretation of how acidification dynamics interact with protein aggregation and ultimately determine the textural and structural properties of the resulting tofu gel.

3.1.2. Texture

The coagulation of soybean protein is a critical process in tofu manufacturing, wherein the soluble proteins in soybean milk undergo denaturation and subsequent aggregation to form a three-dimensional gel network (Guan et al., 2021). The coagulation of tofu using GDL is a process that significantly influences the physicochemical properties and quality of the final product. Table 2 indicates that, with the exception of the ultrasonically treated samples, soybeans derived from germinated soymilk exhibited more hardness than the original beans, with values ranging from 122.0 to 277.3 g and 33.6 to 48.3 g, respectively. Nonetheless, the ultrasonically treated samples exhibited no variation in hardness between the two soybean varieties (30.7–31.7 g). The adhesiveness and springiness indices of all samples exhibited minimal variation, ranging from 8.19 to 10.69 mJ and 0.30 to 1.80 mm, respectively. GDL acts as an acid coagulant, promoting the formation of a uniform and dense gel network in tofu, which is maintained by strong intermolecular forces, particularly hydrophobic interactions (Khoder et al., 2024). The use of GDL as a coagulant also leads to tofu with a high breaking force and penetration distance, indicating a firm texture. In the context of these germinated soymilks and GDL, unfolded and hydrolyzed proteins can form stronger intermolecular bonds with each other, leading to firmer curds. For sonication, ultrasound modifies the rheological properties and microstructure of tofu, leading to a denser and more stable network structure, which is beneficial for tofu texture (Lin et al., 2016). In the meantime, G. Liu et al. (2024) stated that protein hydrolysis during germination generally results in weaker gel networks with reduced firmness, though the extent varies with germination duration and conditions. In addition, this study also reported that heat treatment also increased the firmness of soybean curd. Heat treatment plays a crucial role in enhancing the gel strength of sprouted soybean curd by promoting protein denaturation and improving the interactions between protein components. The denaturation of soy proteins during heat treatment increases their hydrophobicity, facilitating stronger protein-protein interactions that are vital for gel formation. According to (Guan et al., 2021), tofu gel formation is primarily governed by denaturation and aggregation of glycinin (11S) and β-conglycinin (7S). During heating and acidification, these proteins unfold, exposing hydrophobic groups and sulfhydryl residues, which promote intermolecular disulfide bond formation and hydrophobic interactions. These interactions strengthen the gel network and contribute to the mechanical stability of tofu. The revised manuscript now reflects these soy-specific mechanisms. It can also be seen that the use of physical treatments such as microwave and sonication minimally alter the texture of soybean curd compared to heat treatment but accelerate the coagulation process, which holds substantial importance in industrial manufacturing due to the reduction of production time.

Table 2.

Textural and color attributes of soybean curd of different germination time and treatment methods.

Sample	Hardness (g)	Springiness (mm)	Adhesiveness (mJ)	L*	a*	b*
Ca_0	33.6 ± 0.6a	0.33 ± 0.06a	9.93 ± 0.12ab	85.2 ± 1.6a	1.5 ± 0.6a	20.0 ± 1.0a
Ca_2	122.0 ± 4.4b	0.38 ± 0.03a	10.20 ± 0.21ab	85.3 ± 3.6a	1.3 ± 0.6a	22.7 ± 0.6ab
H_0	35.3 ± 0.6c	1.28 ± 0.15b	8.19 ± 0.69c	87.2 ± 3.4ab	−1.0 ± 0.1b	24.1 ± 1.0b
H_2	125.0 ± 3.0b	1.80 ± 0.26c	10.41 ± 0.55a	90.5 ± 4.0ab	−3.5 ± 0.6c	35.7 ± 0.6c
MW_0	48.3 ± 2.3d	0.43 ± 0.06a	9.31 ± 0.30bd	92.4 ± 2.0b	−1.7 ± 0.6b	20.2 ± 1.0b
MW_2	277.3 ± 8.7e	1.63 ± 0.12bc	10.69 ± 0.37a	91.8 ± 3.0b	−6.5 ± 0.6d	37.1 ± 1.1c
S_0	30.7 ± 1.2c	1.23 ± 0.32b	9.06 ± 0.43 cd	89.4 ± 3.4ab	−6.3 ± 0.5d	32.3 ± 1.7d
S_2	31.7 ± 1.2c	0.30 ± 0.10a	9.44 ± 0.56bd	91.6 ± 1.1b	−6.0 ± 0.1d	34.3 ± 0.6 cd

Notes: Data are presented as mean ± standard deviation of three replicates. Values within a column with different letters indicate significant difference at a significance level of 5% (p < 0.05).

Sample abbreviation: X_a; where X is the treatment methods (Ca, H, MW, and S are calcium salt, heat, microwave and sonication treatment) and a is the germination time (0 and 2 days).

3.1.3. Color

The color values  of GDL-coagulated soybean curds treated with different physical methods are presented in Table 2. The color analysis revealed distinct patterns influenced by both germination and physical treatments. L* values (lightness) ranged from 85.2 to 92.4, with physically treated samples (H, MW, S) exhibiting higher brightness compared to calcium-coagulated samples, attributable to more complete protein denaturation forming homogeneous gel networks with enhanced light-scattering properties (Mahmoudi et al., 2007; C. Yang et al., 2023). The a* values demonstrated a consistent shift from positive to negative following germination across all treatment methods, indicating a transition from slight redness toward greenness. This chromatic change reflects carotenoid degradation concurrent with chlorophyll biosynthesis during germination (Fernández-Marín et al., 2017), with ultrasound-treated (S_0, S_2) and germinated microwave samples (MW_2) displaying the most negative a* values (−6.0 to −6.5), suggesting that physical treatments may facilitate the liberation or preservation of green pigments. The b* values (yellowness) increased substantially after germination, particularly in H_2 (35.7) and MW_2 (37.1), which can be attributed to Maillard reaction products formed between reducing sugars accumulated during germination and amino acids under thermal conditions, as well as the release of yellow-colored flavonoids and isoflavones from the cellular matrix during physical processing. Microwave heating enhances the reaction between reducing sugars and amino groups, generating melanoidin compounds that impart brownish coloration to soymilk and tofu, particularly at higher power levels or extended treatment durations (Alkanan et al., 2024). Similarly, sonication reduces the size of colloidal particles, including pigment-protein complexes, altering light scattering properties and consequently affecting perceived color intensity and brightness (Mozafarpour et al., 2022). Tofu with a high moisture content, as well as a smooth surface and white color, which are desirable sensory attributes (Shi et al., 2020).

3.1.4. SDS-page

The SDS-PAGE analysis of soybean curd samples depicted in Fig. 2 revealed distinct protein profiles across different coagulation and pre-treatment methods. All samples exhibited characteristic bands corresponding to the two major soybean storage proteins, namely β-conglycinin (7S globulin) and glycinin (11S globulin). The α’ and α subunits of β-conglycinin were identified in the 70–80 kDa region, while the β subunit appeared at approximately 45–50 kDa (Kang et al., 2023). The acidic (AS) and basic (BS) subunits of glycinin were observed at 35–40 kDa and 20–25 kDa, respectively (Ning et al., 2024; Z. Yang et al., 2025). Additionally, low molecular weight bands in the 10–15 kDa region were detected, potentially representing Kunitz trypsin inhibitor (KTI) and Bowman-Birk inhibitor (BBI), or peptide fragments generated during germination (da Silva et al., 2024; Herwade et al., 2021). A notable observation was the markedly lower band intensity in calcium-coagulated samples (Ca_0 and Ca_2) compared to those subjected to physical treatments including heat (H), microwave (MW), and ultrasound (S). This difference can be attributed to the distinct coagulation mechanisms involved; calcium ions induce protein aggregation through ionic bridging without requiring complete thermal denaturation (Qi et al., 2024), whereas physical treatments promote protein unfolding and enhanced intermolecular interactions, thereby facilitating greater protein incorporation into the curd matrix (Sun et al., 2023).

Fig. 2.

SDS-PAGE of soybean curd of different germination time and treatment methods. Sample abbreviation: X_a; where X is the treatment methods (Ca, H, MW, and S are calcium salt, heat, microwave and sonication treatment) and a is the germination time (0 and 2 days). Notes: M is the standard protein.

Regarding the effect of germination duration, no substantial differences in protein banding patterns were observed between day 0 and day 2 samples within each treatment group. This finding suggests that a two-day germination period was insufficient to induce significant proteolytic degradation detectable under the current electrophoretic conditions. However, a slight increase in band intensity at the 10–15 kDa region was noted in the ultrasound-treated germinated sample (S_2), indicating potential synergistic effects between ultrasonic cavitation and endogenous protease activity activated during germination (Pacheco et al., 2024).

3.1.5. Secondary structure

FTIR spectra of soybean proteins exhibit several characteristic absorption bands, as shown in Fig. 3. Amide I region (1600–1700 cm⁻¹) is the most informative band for protein secondary structure analysis, primarily arising from C O stretching vibrations of peptide bonds; α-helix structures typically appear at 1650–1658 cm⁻¹ while random coil elements generally show bands at 1640–1650 cm⁻¹. In addition, β-sheet structures demonstrate characteristic absorption at 1625–1640 cm⁻¹ and 1670–1695 cm⁻¹ while β-turns exhibit absorption around 1660–1670 cm⁻¹. Amide II region (1510–1580 cm⁻¹) results from N—H bending (60%) and C—N stretching (40%) vibrations, providing complementary structural information. Amide III region (1200–1350 cm⁻¹) arises from complex vibrations involving C—N stretching, N—H bending, and other contributions, useful for confirming secondary structure assignments. Amide A (3300 cm⁻¹) and amide B (3100 cm⁻¹) are associated with N—H stretching vibrations, sensitive to hydrogen bonding environments (Long et al., 2015). Fig. 4 also estimate the percentage of different protein secondary structure of soybean curd samples. Heat treatment or heat generating processes, such as microwave treatment or sonication, diminish the fraction of α-helix while augmenting the proportion of β-sheet structure. It is reported that thermal processing typically results in decreased band intensity at ∼1655 cm⁻¹ (α-helix) with concurrent increases at ∼1630 cm⁻¹ and ∼ 1685 cm⁻¹ (β-sheet structures), quantitatively reflecting the α-helix to β-sheet transition (Yu, 2005). Furthermore, acidification induced by GDL shifts the amide I band profile, with more pronounced β-sheet characteristics emerging at lower pH values as proteins approach their isoelectric point. The secondary structure of soy proteins is also influenced by processing conditions, such as pH and the presence of coagulants like calcium sulfate, which can affect protein folding and aggregation, thereby impacting tofu's texture and firmness (Liu, Yang, et al., 2024). As shown in Fig. 4, β-sheet was the predominant secondary structure in all soybean curd samples, reflecting the inherently ordered conformation of major soy storage proteins such as glycinin and β-conglycinin (Guan et al., 2021). Notably, microwave and sonication treatments resulted in a markedly higher β-sheet content compared to calcium salt and heat treatments, suggesting that these physical treatments promoted stronger intermolecular interactions and a more ordered protein network during coagulation. The elevated β-sheet content in microwave and sonication treatments is likely attributable to the energy input from these physical fields, which promotes protein unfolding and subsequent intermolecular β-sheet formation through enhanced hydrophobic exposure and hydrogen bonding during gel network assembly.

Fig. 3.

FTIR spectra with focused bands of amide I of soybean curd of different germination time and treatment methods. Sample abbreviation: X_a; where X is the treatment methods (Ca, H, MW, and S are calcium salt, heat, microwave and sonication treatment) and a is the germination time (0 and 2 days).

Fig. 4.

Secondary structure contents of protein in soybean curd of different germination time and treatment methods. Sample abbreviation: X_a; where X is the treatment methods (Ca, H, MW, and S are calcium salt, heat, microwave and sonication treatment) and a is the germination time (0 and 2 days).

Regarding α-helix, an opposite trend was observed between treatment types upon germination: calcium salt and heat-treated samples exhibited a decrease in α-helical content after germination, whereas microwave and sonication-treated samples showed an increase. The divergent response of α-helix to germination between treatment types may reflect differences in how each treatment conditions the protein matrix: heat and calcium salt treatments stabilize a more compact structure that germination partially disrupts, whereas the looser protein conformation induced by microwave and sonication may allow germination-derived peptides or partially hydrolyzed subunits to adopt helical arrangements more readily. Furthermore, germination consistently increased the proportion of random coil across all treatments, indicating a general loosening of protein conformation in germinated soymilk. The universal increase in random coil upon germination is consistent with enzymatic proteolysis during the germination process, where protease activity degrades ordered protein domains and generates flexible, disordered polypeptide segments that persist through coagulation (Pacheco et al., 2024).

3.1.6. Correlation between secondary structure and hardness

The substantially higher hardness observed in germinated samples compared to their non-germinated counterparts, particularly in microwave treatment (MW_2: 277.3 g vs. MW_0: 48.3 g), can be mechanistically explained through the combined effects of germination-induced protein modification and treatment-specific structural reorganization. Germination activates endogenous proteases that partially hydrolyze storage proteins, exposing buried hydrophobic regions and increasing the availability of reactive sites for intermolecular interactions during coagulation (Pacheco et al., 2024). This pre-treatment of the protein matrix, when subsequently subjected to physical treatments, facilitates more extensive protein cross-linking and network formation, ultimately yielding a firmer gel structure. The relationship between β-sheet content and hardness is most clearly demonstrated in the microwave-treated samples. MW_2 exhibited the highest β-sheet proportion among all treatments (62.7%), coinciding with the highest hardness value recorded (277.3 g), whereas MW_0, despite sharing the same treatment modality, showed comparably high β-sheet content (77.4%) yet substantially lower hardness (48.3 g). This apparent paradox suggests that β-sheet content alone does not determine gel hardness; rather, the origin and arrangement of these structures matter. In MW_0, the high β-sheet content likely reflects a tightly pre-aggregated protein network formed directly under microwave heating, whereas in MW_2, germination-induced partial unfolding creates a more flexible precursor that, upon microwave treatment, reaggregates into a denser and more mechanically robust cross-linked matrix. β-sheet structures are characterized by extended polypeptide chains stabilized through extensive intermolecular hydrogen bonding networks. Unlike α-helical conformations that are predominantly intramolecular and compact, β-sheets facilitate the formation of robust cross-links between adjacent protein molecules, thereby creating a more rigid and interconnected gel matrix (Thurber et al., 2024). The microwave treatment induces rapid and volumetric heating that promotes protein unfolding and subsequent reaggregation into β-sheet-rich configurations. This rapid thermal processing minimizes localized overheating while maximizing uniform energy distribution, favoring the alignment of polypeptide chains into ordered β-sheet assemblies rather than disordered aggregates (Rombouts et al., 2020). The elevated β-turn content in MW_2 (13.7%) further supports the formation of complex antiparallel β-sheet architectures, which provide additional structural reinforcement to the curd.

In contrast, the negligible hardness difference between S_0 and S_2 (30.7 vs. 31.7 g) reveals that ultrasonication effectively counteracts the textural enhancement typically induced by germination. Despite germination increasing random coil content in S_2, the cavitational forces generated during ultrasonic treatment fragment protein aggregates and partially hydrolyzed peptides into smaller units incapable of forming extensive cross-linked structures. The FTIR data supports this interpretation, as S_2 exhibited a markedly lower β-sheet content (57.8%) relative to MW_2, alongside higher random coil proportion, indicating that ultrasonic cavitation suppresses the reorganization of unfolded proteins into ordered β-sheet networks essential for gel strength development. Consequently, regardless of germination-induced biochemical changes, the physical disruption imposed by ultrasonication dominates the final textural outcome, resulting in uniformly soft curds across both germination timepoints.

3.2. Characterization of soybean whey separated from soybean curd

3.2.1. pH, TSS and turbidity

Table 3 reveals the physicochemical parameters of whey derived from soybean curd coagulated with GDL in conjunction with several physical treatments. The soluble solid content of the samples exhibited minimal variation, with values between 1.8 and 3.0 (°Brix). Conversely, the pH of the physically treated samples was lower than that of the control sample with calcium salt, ranging from 4.77 to 5.79, with the ultrasonically treated sample exhibiting the lowest pH at 4.77. In addition, the turbidity of the examined samples (2.2–11.5 FNU) was markedly lower than that of the control sample (27.5–28.4 FNU). The more turbid whey observed in the control sample using calcium salt as coagulant may be explained by the excessive calcium ion migrated from curd to soybean whey. Calcium ions neutralize surface charges on colloidal particles in soymilk, destabilizing the suspension and promoting aggregation of both protein and non-protein components (Zamani et al., 2020), leading to higher turbidity.

Table 3.

Some selected physicochemical of soybean whey separated from soybean curd of different germination time and treatment methods.

Sample	pH	TSS (°Brix)	Turbidity (FNU)
Ca_0	5.65 ± 0.02a	2.0 ± 0.1a	28.4 ± 2.1a
Ca_2	5.79 ± 0.07a	2.8 ± 0.1b	27.5 ± 2.2a
H_0	5.22 ± 0.04bc	1.8 ± 0.1c	2.2 ± 0.3b
H_2	5.56 ± 0.05ab	2.4 ± 0.1d	2.1 ± 0.2b
MW_0	5.17 ± 0.11c	2.0 ± 0.1a	4.1 ± 0.1c
MW_2	5.49 ± 0.12abc	3.0 ± 0.2e	4.4 ± 0.2c
S_0	4.77 ± 0.29d	2.0 ± 0.1a	10.3 ± 0.8d
S_2	4.77 ± 0.32d	2.0 ± 0.1a	11.5 ± 0.6d

Notes: Data are presented as mean ± standard deviation of three replicates. Values within a column with different letters indicate significant difference at a significance level of 5% (p < 0.05).

Sample abbreviation: X_a; where X is the treatment methods (Ca, H, MW, and S are calcium salt, heat, microwave and sonication treatment) and a is the germination time (0 and 2 days).

3.2.2. Phenolic and FRAP activity

Fig. 5A illustrates the phenolic content and FRAP iron-reducing activity of whey extracted from soybean curd subjected to various treatments. With regard to phenolic content, all whey samples derived from germinated seeds exhibited greater phenolic levels than the non-germinated samples, except for the steaming-treated samples. Moreover, microwave treatment yielded whey with the greatest phenolic content (351.9–371.2 mg GAE/L), followed by Ca samples, S samples, and H samples. Nonetheless, there was no notable disparity in FRAP activity between the non-germinated and germinated samples, with the exception of the H_2 sample, which exhibited a greater FRAP value (77.8 mg TE/L) in comparison to the H_0 sample (34.4 mg TE/L). It can be concluded that the FRAP antioxidant activity was independent of the total phenolic content in soybean whey. Germination may generate specific phenolic compounds with varying reducing capacities, and physical treatments differentially affect the stability and reactivity of these individual constituents (Shabbir et al., 2022). Germination of soybeans has been shown to significantly increase the content of gamma-aminobutyric acid (GABA), isoflavones, and total polyphenols, which are beneficial for health and enhance the sensory qualities of soy products (Zhao et al., 2021). Additionally, potential interactions between phenolics and proteins or other matrix components in whey may modulate their electron-donating ability in the FRAP assay. The FRAP values obtained in this study were lower than those reported for soybean whey in some previous studies. This discrepancy can be attributed to differences in raw material characteristics (germinated vs. non-germinated soybeans), processing conditions, and analytical protocols. Several studies reporting higher FRAP values typically employed concentrated whey fractions, different extraction/dilution schemes, or alternative antioxidant assays calibrated under different reaction conditions.

Fig. 5.

(A) Antioxidant properties and (B) amino acid profiles of soybean whey separated from soybean curd of different germination time and treatment methods. Sample abbreviation: X_a; where X is the treatment methods (Ca, H, MW, and S are calcium salt, heat, microwave and sonication treatment) and a is the germination time (0 and 2 days).

The inverse relationship between FRAP activity and phenolic content in germinated samples treated with microwave and ultrasound (r = −0.86 and − 0.99, respectively) as opposed to positive correlation in Ca samples (r = 0.99) can be explained through the structural characteristics of the resulting curd. As demonstrated by texture analysis, MW_2 produced the hardest curd (277.3 g) with a dense, highly cross-linked protein network stabilized by extensive β-sheet structures (41.14%). This compact gel matrix effectively entraps phenolic compounds within the curd through hydrogen bonding and hydrophobic interactions with aggregated proteins, limiting their partitioning into the whey fraction. Similarly, although S_2 exhibited softer texture, the ultrasound-induced protein-phenolic complexation through enhanced molecular interactions during cavitation retained antioxidant compounds within the curd structure. Consequently, despite germination increasing total phenolic biosynthesis, the physical treatments that promote stronger protein aggregation and gel network formation simultaneously reduce phenolic migration into whey, resulting in lower measured FRAP activity. This structural retention mechanism explains why whey from firmer, more structured curds exhibits diminished antioxidant capacity despite originating from phenolic-enriched germinated soymilk. Microwave heating promotes Maillard reactions between reducing sugars and amino compounds, generating melanoidins and other reaction products with novel antioxidant properties (Xiang et al., 2020). Rapid thermal inactivation of oxidative enzymes (polyphenol oxidase, lipoxygenase) may preserve native antioxidant compounds that would otherwise be enzymatically degraded during processing (Cavalcante et al., 2021). For sonication treatment, cavitation phenomena during sonication disrupt cellular structures and weaken bonds between phenolic compounds and macromolecules, potentially increasing the concentration of free phenolics in the whey fraction (Gajendragadkar & Gogate, 2016). Sonication can promote the conversion of isoflavone glycosides (daidzin, genistin) to their more bioactive aglycone forms (daidzein, genistein) through localized energy transfer, potentially enhancing antioxidant capacity (Fahmi et al., 2014). Sonication alters protein conformation, potentially exposing previously buried amino acid residues with antioxidant properties (e.g., sulfur-containing amino acids, aromatic residues). Additionally, ultrasound treatment can increase the content of bioactive compounds such as isoflavones in soymilk, which may influence the nutritional profile of the tofu (Lee & Lee, 2023).

3.2.3. Amino acid profiles

The amino acid profiles of soybean whey depicted in Fig. 5B indicate that germination markedly elevated several amino acids in soybean seeds, particularly Glu (from 574 to 901 to 1901–1972 mg/L, 2.1–3.4 times more), Asp (from 350 to 506 to 862–1015 mg/L, 2.4–2.9 times greater), and Arg. This substantial Glu enrichment during germination can be attributed to enhanced proteolytic activity and the conversion of glutamine to glutamate by activated glutaminase enzymes (Bera et al., 2023). The elevated Glu content presents significant valorization potential, as this amino acid serves as a natural flavor enhancer with umami taste properties, suggesting potential applications in savory food formulations. Furthermore, various amino acids like Pro, Ala, Ser, and Thr exhibited a rise in the whey samples derived from germinated soybeans. In comparing the amino acid profiles of the samples with different physical treatments, it is evident that physical treatment released amino acids more effectively from the soybean curd structure, particularly in the germinated soybean curd, than calcium salt treatment, as indicated by the comparable or reduced levels of numerous amino acids in sample Ca_2 relative to Ca_0. It is reported that the rapid heating achieved through microwave energy promotes D/L-amino acid racemization, particularly affecting aspartic acid and serine, potentially altering nutritional bioavailability (Aristoy & Toldrá, 2021). In the meantime, cavitation forces during sonication promote protein hydrolysis, increasing the concentration of free amino acids and small peptides in the whey fraction. Sonication alters protein tertiary structure, potentially affecting the extractability and subsequent detection of specific amino acids during analysis (Rahman & Lamsal, 2021).

4. Conclusions

Germinated soybeans have significant promise for tofu and soybean whey production systems due to their unique biochemical composition. The integration of thermal and non-thermal processing methods, including microwave and sonication, presents a viable alternative to calcium salt coagulation in practical manufacturing. These changes provide soymilk with distinctive coagulation kinetics, generally marked by accelerated gelation rates. The accelerated coagulation achieved through physical pre-treatments presents significant implications for industrial tofu production. Reduced coagulation time directly translates to enhanced production throughput and energy efficiency, potentially enabling manufacturers to increase batch cycles without additional capital investment. In terms of soybean whey, which is a by-product of tofu production, there is potential for its valorization through various methods, including the recovery of nutrients and the production of functional ingredients, although challenges remain in minimizing waste generation. The whey fraction derived from germinated soybeans demonstrates elevated levels of free amino acids, and antioxidant compounds compared to conventional counterparts. The integration of germinated soybeans into tofu production not only enhances the nutritional and sensory qualities of the final product but also aligns with sustainable practices by potentially reducing waste and improving the valorization of by-products like soy whey. Furthermore, the separated soybean whey, traditionally regarded as a waste stream, represents an underutilized resource rich in free amino acids, particularly glutamic acid released during germination and coagulation processes. This glutamate-enriched whey could be valorized as a natural flavor enhancer for savory food applications or as a nitrogen source for microbial fermentation. Additionally, the residual oligosaccharides and bioactive peptides in whey offer opportunities for developing functional food ingredients or nutraceutical products, thereby contributing to a zero-waste biorefinery approach in soybean processing. Despite its efficacy in enhancing curd texture, microwave processing presents scalability challenges for industrial applications. The inherent non-uniformity of electromagnetic field distribution can create localized hotspots during bulk processing, potentially causing inconsistent protein denaturation and heterogeneous gel formation. Optimization of cavity design, turntable mechanisms, and power modulation protocols is therefore essential for ensuring reproducible product quality at production scale.

Author contribution

Thi-Van-Linh Nguyen: Conceptualization; Data curation; Investigation; Methodology; Writing - original draft. Thi Tuong Vi Tran: Data curation; Investigation; Methodology; Writing - original draft. Quoc-Trung Huynh: Data curation; Investigation; Writing - original draft. Thanh-Thuy Dang: Data curation; Writing - original draft. Thi-Thuy-Dung Nguyen: Data curation; Writing - original draft. Phuoc-Bao-Duy Nguyen: Data curation; Visualization. Anh Duy Do: Data curation; Investigation. Quoc-Duy Nguyen: Conceptualization; Data curation; Investigation; Visualization; Methodology; Writing – original draft; Writing - review & editing.

CRediT authorship contribution statement

Thi-Van-Linh Nguyen: Writing – original draft, Methodology, Investigation, Data curation, Conceptualization. Thi Tuong Vi Tran: Writing – original draft, Methodology, Investigation, Data curation. Quoc-Trung Huynh: Writing – original draft, Investigation, Data curation. Thanh-Thuy Dang: Writing – original draft, Data curation. Thi-Thuy-Dung Nguyen: Writing – original draft, Data curation. Phuoc-Bao-Duy Nguyen: Visualization, Data curation. Anh Duy Do: Investigation, Data curation. Quoc-Duy Nguyen: Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation, Data curation, Conceptualization.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.