A Green Workflow to Determine Flavonoids from Physalis angulata L.: Extraction Optimization by Response Surface Method and Spectrophotometric Method Validation

Abstract

Wild Physalis angulata L. has promising medicinal potential due to its rich flavonoids. However, a green analytical approach for these compounds from this plant has not yet been thoroughly optimized. Therefore, this study optimized ultrasound-assisted extraction using the response surface method for the UV-VIS spectroscopic determination of the total flavonoid content in P. angulata in Vietnam. Notably, the greenness of the whole procedure was evaluated by AGREE, Eco-Scale, GAPI, BAGI methodologies. The Box–Behnken model was applied to design the experiments with four variables: ethanol concentration, solid-to-liquid ratio, extraction temperature, and time. The UV-Vis spectrophotometric method was validated at 510 nm according to AOAC guidelines and met all the requirements, including specificity, linearity (R² = 0.9996) in the working range of 15–120 µg/mL, repeatability (RSD = 1.89%), intermediate precision (RSD = 2.21%), and accuracy (recoveries from 99.52 to 104.06%). The limits of detection (LOD) and quantification (LOQ) were 2.48 µg/mL and 7.52 µg/mL, respectively; however, to avoid noise signal at lower concentrations, the validated lower limit of quantification (LLOQ) was set at 15 µg/mL. Data were analyzed using second-order regression. The R² = 0.9726 shows a close correlation between variables and the experimental data. The optimal extraction conditions were 31.66% ethanol, 30:1 mL/g ratio, 80 °C and 48.73 min. The predicted values (38.09 ± 1.70 mg RU/g) were not statistically different from the experimental values (34.58 ± 0.87 mg RU/g), confirming the model’s accuracy and applicability in optimizing the extraction process. The ultrasound-assisted extraction was optimized to enhance the flavonoid extraction yield from P. angulata, providing a solid scientific foundation for further pharmacological research.

1. Introduction

The study of the chemical composition and nutritional value of plants has drawn much attention due to their applications and value in the development of new medicines. Physalis angulata L., an eminent member of the Solanaceae family, has historically been endemic to the tropical regions of the Americas, but later, it has spread across numerous tropical regions, notably Asia and Africa. In Vietnam, P. angulata is a wild plant often encountered in grasslands across the country; it has long been consumed as a vegetable in daily meals, and has also been used as a folk remedy for various diseases [1]. The stems, leaves, fruits, and calyxes of the genus Physalis are the primary medicinal components employed in the treatment of inflammatory diseases. Their stems and leaves have been employed in traditional medicine to treat dermatitis, malaria, bronchitis, hepatitis, and rheumatism in China, Mexico, Indonesia, Peru, and Brazil [2]. The entire plant of P. minima has been widely used to treat gout, urinary disorders, ascites, and bladder ulcers [3]. The fruits and calyxes of P. alkekengi var. franchetii (Mast.) Makino, also referred to as “Jin Denglong” within the Chinese tradition, have been utilized in both dietary supplements and conventional medicine in China. Their therapeutic applications include treatments for ailments such as sore throat discomfort, cough, and painful urination [4]. The use of extracts and preparations of P. angulata, including crude extract and decoction, in the treatment of diseases such as malaria, asthma and dermatitis has been recorded [5].

Flavonoids have also been associated with a variety of biological activities, including antibacterial, anti-inflammatory, antioxidant, antidiabetic, neuroprotective and anticancer activities [5,6,7,8]. They are common secondary metabolites in plants with a characteristic structure consisting of two benzene rings (rings A and B) and a pyrene ring (ring C) linked between the two rings. The C6-C3-C6 backbone structure of flavonoids allows for many modifications of hydroxyl, methoxy, and glycosyl groups, resulting in a variety of different flavonoids such as flavones, flavonols, flavanones, isoflavones, anthocyanidins, and chalcones [9]. Some flavonols, such as quercetin, have the ability to scavenge free radicals and inhibit cancer cells [7,8], and also contribute to antidiabetic effects [10]. Their hydroxyl groups form stable complexes in acid with aluminium chloride, allowing them to be quantified completely by UV-Vis. Compared with the Folin-Ciocalteau method for phenolic acids [7], aluminium chloride is more specific for flavonoids, which is favourable for further studies on the biological and pharmacological activities of only this group. Moreover, the availability and low cost of standards (rutin, quercetin) for calibration make the workflow more eco-efficient.

Due to the important roles of flavonoids in pharmaceutical and industrial applications, many studies have investigated extraction techniques to obtain the highest flavonoid content extracted from natural materials. Accordingly, many different extraction methods have been proposed, including maceration, percolation, steam distillation, boiling, reflux, cold maceration, and Soxhlet extraction [11]. Although these methods provide high extraction efficiency, they consume a lot of solvents that are hazardous to the operators and the surrounding environment. Modern extraction techniques have been developed to overcome the limitations of traditional methods, including ultrasound-assisted extraction (UAE) [12], supercritical fluid extraction (SFE) [13], microwave-assisted extraction (MAE) [14], solid-phase extraction (SPE), enzyme-assisted extraction, pressurized liquid extraction (PLE or accelerated solvent extraction-ASE), pulsed electric field (PEF)-assisted extraction, as well as multi-technique combination systems [15]. These advanced methods have demonstrated high extraction efficiency from various natural sample matrices in a short time, in particular, significantly reducing the use of toxic organic solvents through the application of environmentally friendly “green” solvents, in line with the trend of sustainable development [16].

Among the green extraction methods proposed above, ultrasound-assisted extraction is a modern but easy to use, environmentally friendly and efficient technique to not only extract large amounts of flavonoids in a short time but also preserve active constituents and reduce costs, thereby being chosen in this study. Flavonoids are susceptible to temperature effects, and their extracted contents depend on the extraction time and solvent, as well as its ratio to the material. This provides a rationale for optimizing these parameters to obtain the highest flavonoid content for the investigation of therapeutic actions of flavonoids. In this regard, modern tools available for laboratory planning, particularly response surface methodology (RSM), have been used for this purpose in the literature, and especially for the extraction of polyphenols from Physalis angulata L. [17,18]. With the aforementioned reasons, along with the paucity of research documents on this issue, this study was carried out to optimize these conditions, which were then applied to develop a comprehensive UV-Vis spectroscopic procedure to determine the total flavonoid content extracted from leaves of Physalis angulata L. collected from different regions throughout Vietnam. Another important aspect of phytochemical research is green analytical chemistry (GAC). Its core importance exhibits environmental sustainability initiatives to eliminate or reduce the harmful impacts of chemical processes, operational costs, time and resources. Adhering to the twelve principles of GAC, the research employed four methods to assess the greenness of the whole workflow, including Eco-Scale [19], Green Analytical Procedure Index (GAPI) [20,21], Analytical GREEnness Metric Approach and Software (AGREE) [22], and Blue Applicability Grade Index (BAGI) [23].

2. Materials and Methods

2.1. Sampling and Pre-Treatment

The whole plants at mature green with fruits were collected in the morning under dry weather conditions to minimize seasonal variation from four regions throughout Vietnam: Tien Giang, Nam Đinh, An Giang, and Can Gio from January to March 2025 (a total of 5 kg for each region). The plants were authenticated by plant scientists at the Institute of Environmental Science, Engineering and Management, Industrial University of Ho Chi Minh City, and referred their plant morphology to documents [24] to identify them as Physalis angulata as follows. The herb is 10–100 cm tall. It is hairless or with a few short appressed hairs. The stems are angled and hollow, with lower branches that sometimes root at the nodes. The leaves are 4–15 cm × 2.5–10 cm. The margins are irregularly serrated or entire, with short hairs, and the petiole is 2–11 cm long. The flowers are solitary; the sepals are 3–5 mm long when in inflorescence, 2–4 cm long when fruiting, and yellow-green with purple veins. The petals are 5–10 mm long, up to 1 cm in diameter, pale yellow, with or without dark spots, and have a triangular spot of dense short hairs at the throat. The anthers are entirely pale blue. The berries are yellow, 10–16 mm in diameter.

Roots, stems, leaves and fruits were separated from the whole plants, thoroughly washed with tap water to remove all the soil particles, classified, and air-dried. The dried materials were ground into fine powder by a milling machine, sieved through a 0.710 mm sieve and stored in paper bags at 22 °C until use.

2.2. Single-Factor Investigation of Flavonoid Extraction

Methanol and hexane have been used for the extraction, but they are not green solvents because they are toxic, carcinogenic and hazardous to the environment. Ethanol, a polar solvent, has been proven effective for extracting bioactive substances, including phenolic compounds, and gives a safe extract for further use in the production of food and pharmaceuticals [25]. Indeed, the preliminary extraction showed higher efficiency than methanol. In addition, for the green purpose, ethanol was chosen to investigate the best concentration for the ultrasound-assisted extraction.

The preliminary test was as follows: a total of 0.50 g of each powdered plant part was extracted with ethanol on a sonication water bath (DaiHan WUC-D10H, Republic of Korea) under constant ultrasonic power (60 kHz). The extraction starts up with the room temperature (30 °C), and other conditions were selected based on the literature: 30 min [26], ethanol 70% [27], material-to-solvent ratio (MSR) of 1:25 g/mL in order to choose the part that gave the highest total flavonoids.

Separate experiments were conducted to choose the level producing the highest total flavonoid content (TFC) among other levels of extraction temperature (30, 45, 60, 75 °C), time (15, 30, 45, 60 min), concentration of ethanol (25%, 50%, 70%, 96%) and MSR (1:10, 1:25, 1:40, 1:55 g/mL) to put into the SRM model (Figure 1). The extract was evaporated to remove the solvent and dried to a constant mass at 110 °C. The remaining residues were stored at 5 °C for subsequent analyses.

In addition, this study investigated dechlorophyllization to remove chlorophyll, which is the most important influencing factor for the UV-Vis determination of total flavonoids in leaves. Two proposed methods are liquid–liquid extraction and sedimentation. The procedures were based on the study of Betel (Piper betle L.) concerning leaf ethanolic extract [27], but slightly modified to suit the actual condition. The ultrasonic extraction with 70% ethanol (the MSR of 1:25 g/mL) was at 30 °C for 30 min. For liquid–liquid extraction, the ethanolic extract was shaken with hexane (ratio 1:1, v/v), repeated 3 times and the lower solvent layer was recovered. For sedimentation, after removing ethanol from the extract, distilled water was added at a ratio of 1:1 (v/v) to proceed with the sedimentation at 4 °C for 24 h. The mixture was then centrifuged at 10,000× g for 30 min at 4 °C to separate the supernatant for further analysis. The initial and dechlorophyllized samples were measured spectrophotometrically at 660 nm to evaluate the chlorophyll removal efficiency, based on the method of Lichtenthaler and Buschmann (2001) [28]. At the same time, their TFCs were determined to compare the loss of flavonoids of both methods.

2.3. Total Flavonoid Extraction Optimization by Response Surface Methodology

A four-factor Box–Behnken model was designed to optimize extraction parameters, including ethanol concentration (25–96% v/v), sonication time (15–60 min), extraction temperature (30–75 °C), and ratio (1:10–1:55 mL/g), to maximize TFC (mg Rutin/g dried material) (Figure 1). The experiment consisted of 27 runs, with three replicates at the centre points to assess repeatability. Each extract was analyzed in triplicate to ensure accuracy and reliability. Data was modelled using quadratic regression analysis to establish the relationship between TFC and independent variables. A one-way analysis of variance (ANOVA) was performed using IBM SPSS Statistics (version 27; IBM, 2020) to assess the statistical significance of the model, with a 95% confidence interval (CI). Response surface plots were generated on the software Design-Expert (version 13.0.5.0, Stat-Ease Inc, USA, 2022) to visually represent the interactions between independent and dependent variables. A p-value of less than 0.05 was considered statistically significant, confirming the reliability of the model.

2.4. Development and Application of UV-Vis Spectroscopic Method for Total Flavonoid Contents

The procedure was developed based on the study of Christ (1960) [29]. TFC was determined spectrophotometrically using rutin (87.1% purity, provided by the Institute for Drug Quality Control, Ho Chi Minh City, Lot: QT152 0823) as a standard. The rutin stock solution was prepared at 2000 µg/mL, diluted to 100 or 50 or 40 µg/mL. The method was validated according to AOAC (2023) [30] on a Shimadzu UV-Vis 1900i spectrophotometer (Japan):

-

Specificity: the spectra of blank, sample (after being extracted with the optimized condition), standard and spiked standard were scanned from 450 to 800 nm.

-

Linearity: stock standard (2000 µg/mL) was diluted into standards from 15 to 120 µg/mL. F- and t-tests were applied to confirm the suitability of the regression equation and the significance of the regression coefficient, respectively.

-

Precision: 6 test samples were repeated independently with the same optimized extraction and spectroscopic workflow in one day to calculate RSD (intra-day precision). For inter-day precision, an additional 6 samples were processed on the second day. F-test (two-sample for variances) was used to assess the homogeneity of variance (RSD) between two days, and the t-test (two-sample assuming equal variances) was used to compare the mean of TFC quantified on two different days.

-

Accuracy: the validation was performed on spiked standards at 80%, 100% and 120% levels to calculate recoveries, whether they fell in the range of 95% to 105% (RSD ≤ 3.7%).

-

Limit of detection (LOD) and limit of quantification (LOQ) were calculated from the standard curve with the following equation: LOD = 3.3$\sigma$/S, LOQ = 10$\sigma$/S.

For spectroscopic assay, 1 mL of sample was added to a 10 mL volumetric flask, followed by the addition of 4 mL of distilled water and 0.3 mL of 5% NaNO₂ solution. After 5 min, 0.3 mL of 10% AlCl₃ solution was added. After an additional 6 min, 2 mL of NaOH 1 M was added, and the solution was diluted to volume with distilled water. The mixture was incubated at room temperature for 10 min. A blank was prepared in the same manner as the sample.

The validated method was then applied to determine TFC in actual samples. The remaining residues were dissolved in distilled water in a 20 mL volumetric flask. 1 mL of the sample solution was transferred into a 10 mL volumetric flask for the spectroscopic assay. The TFC was calculated as rutin equivalents based on the calibration curve using the following formula:

$$\mathrm{T}\mathrm{F}\mathrm{C}\left(\mathrm{m}\mathrm{g} \mathrm{R}\mathrm{U}/\mathrm{g}\right)={\displaystyle \frac{\mathrm{C}\times \mathrm{K}\times \mathrm{V}}{1000\times \mathrm{m}\times (1-\mathrm{w})}}$$

where C: flavonoid concentration from the standard curve (µg/mL); K: dilution factor; m: sample weight (g); w: moisture content of the material (%).

3. Results and Discussion

3.1. Results of Single-Factor Investigations Affecting the Flavonoid Extraction

The preliminary testing of powdered leaves provides initial information on the materials: moisture (8.58%), total ash (14.55%), acid-insoluble ash (0.61%), and a relatively high extractive content of 35.9%, indicating a high potential for recovering target compounds during the extraction.

The initial extraction shows that leaves contained the highest TFC (21.64 ± 0.18 mg RU/g), followed by the fruits (7.57 ± 0.23 mg RU/g), while roots and stems had very low TFC with no statistically significant difference (<0.76 mg RU/g, p < 0.05). This agrees with Pillai et al. (2022) (TFC of 370.64 ± 4.33 mg QE/g in leaves vs. 130.48 ± 2.60 mg QE/g in fruit) in Physalis angulata [8]. Thus, leaves were identified as the richest source of total flavonoids and, given their abundant biomass and year-round availability, are considered the most suitable part for further studies.

In terms of dechlorophyllization (

Table 1. Chlorophyll removal results.

Chlorophyll Removal Methods	Chlorophyll Removal Yields (%)	TFC (mg RU/g)
		Before Removal	After Removal
Liquid–liquid extraction with hexane	32.00 ± 7.30	18.25 ± 0.61 *	19.17 ± 0.17 *
Sedimentation	94.46 ± 0.52	18.93 ± 1.26 **	9.23 ± 1.36 **

* No statistically significant difference (p > 0.05); ** Statistically significant difference (p < 0.05).

), liquid–liquid extraction with hexane removed only 32.00 ± 7.30 of chlorophyll but gave higher absorbance values after chlorophyll removal (p > 0.05). However, these higher values were not completely attributed to the flavonoids, since the residual chlorophyll could produce the absorbance in the wavelength region of 400–700 nm. The sedimentation method achieved extremely high chlorophyll removal efficiency (94.46 ± 0.52) but significantly reduced TFC after treatment (p < 0.05), from 18.93 ± 1.26 to 9.23 ± 1.36 mg RU/g, likely due to the precipitation of poorly water-soluble flavonoids such as rutin in cold treatment and centrifugation when ethanol was replaced with water during reconstitution. To overcome this flavonoid loss, the supernatant was lyophilized and stored in the freezer before analysis [27]. Comparing the two methods, the sedimentation is advantageous due to its high chlorophyll removal efficiency, good reproducibility, suitability, more importantly, less time, labor and solvent consumption for the daily analysis of many batches.

The extraction yields depended on both the viscosity and polarity of solvents, as well as the polarity of the compounds. Solvent concentration had a significant effect on TFC (p < 0.001). It was found that an increase in the ethanol ratio from 25 to 96% resulted in a decrease in flavonoid content. The highest TFC was obtained with 25% ethanol (17.65 ± 0.67 mg RU/g), compared to 50% ethanol (15.37 ± 0.73 mg RU/g) (Figure 2a). 25% ethanol has higher polarity than 50% ethanol, while some flavonoids are nonpolar or less polar, such as quercetin. The high polarity of 25% ethanol may extract polar phenolic acids as well. Therefore, it is essential to optimize this factor, and 50% ethanol was chosen instead of 25% to put into the RSM model.

Ultrasonic-assisted extraction was performed at 30 °C, with the ratio of 1 g of powdered leaves to 25 mL of 50% ethanol (g/mL) in 15, 30, 45, and 60 min. The statistical analysis shows the significant influence of extraction time on TFC (p < 0.001). TFC increased markedly from 12.56 ± 0.71 mg RU/g (15 min) to 16.27 ± 0.45 mg RU/g (45 min), indicating a substantial improvement in extraction efficiency during the first 45 min. Between 45 and 60 min, TFC was almost unchanged (Figure 2b), with no statistically significant difference (p > 0.05). There was a slight signal of reduced TFC and reduced reproducibility in longer extraction time, which was observed by a higher standard deviation at 60 min (16.28 ± 1.04 mg RU/g). Therefore, the sonication time of 45 min was adequate to save time, avoid the degradation of phenolic compounds [31], so ensure the extraction efficiency and experimental consistency.

Active compounds concentrate in unknown regions of the materials, limiting their movement outside the cell for the extraction [32]. The flavonoid extraction can be accelerated using ultrasonic energy, but excessive energy can form free radicals and thus damage bioactive compounds [12,15]. A higher volume of solvent can dissolve more analytes, enhancing the solubility and diffusion of flavonoids. So, solid-to-liquid ratios (from 1:10 to 1:55 g/mL) are investigated using 50% ethanol at 60 °C for 45 min. It was observed that the flavonoid contents increased with an increase in the volume of the extracting solvent from 10 to 40 mL. The statistical analysis indeed shows the effects (p < 0.001). The 1:10 ratio yielded the lowest TFC (17.81 ± 0.12 mg RU/g) (Figure 2c). Higher ratios (1:25, 1:40, and 1:55) resulted in significantly increased TFC, but with no statistical difference among them. This reveals the saturation of the extraction system. The higher ratios are not really meaningful to maximize the TFC yields. Thus, the 1:25 ratio was identified as optimal to ensure high extraction efficiency with economic and operational feasibility; this selection is also near the ratio (1:30 g/mL) used in previous studies on different plants [33].

Finally, temperature is one of the main factors affecting the degradation of bioactive compounds. The flavonoids were extracted with 50% ethanol for 45 min at a solid-to-liquid ratio of 1:25 (g/mL) at 30 °C, 45 °C, 60 °C, and 75 °C, respectively. The results showed a significant influence of temperature on extraction efficiency (p = 0.003). TFC increased from 14.86 ± 1.15 at 30 °C to 16.01 ± 1.07 mg RU/g at 45 °C (Figure 2d), although the difference was not statistically significant (p > 0.05). At 60 °C and 75 °C, TFC increased significantly to 18.71 ± 1.02 and 19.12 ± 0.95 mg RU/g, respectively. Increasing temperatures significantly enhance extraction efficiency due to the increase in cavitation bubbles, contact area, material porosity, and solvation, resulting in increased mass transfer and diffusion between the solid and solvent [15]. However, too high a temperature could not only increase the extraction but also likely degrade the compounds, such as quercetin, especially at 100 °C [34]. This is the reason why rutin was used as a standard in our study [35]. Thus, 60 °C is finally chosen.

3.2. Extraction Optimization Results by Response Surface Methodology

The optimal value of all four factors was selected as the central level (level 0) in the optimization experiment design. The corresponding experimental ranges of each factor are shown in

Table 2. Experimental optimization design.

Independent Variables	Abbreviations		Value Levels
	Mark	Code	Min (−1)	Centre (0)	Max (+1)
Ethanol concentration (%)	C	X1	25	50	75
Sonication time (minute)	T	X2	20	45	70
Extraction temperature (°C)	T	X3	40	60	80
Solvent-to-material ratio (mL/g)	D	X4	20	25	30
Dependent variable			Requirement
Total flavonoid contents (mg RU/g)	TFC	Y	Maximum

.

Twenty-seven experiments were designed using the Box–Behnken model. The corresponding flavonoid contents obtained from P. angulata leaves under each condition are presented in

Table 3. Experimental data for Box–Behnken model.

No	Samples	Code	Independent Variables				Dependent Variables
			X1 (%)	X2 (min)	X3 (°C)	X4 (mL/g)	TFC (mg RU/g)
1	A1	−−00	25	20	60	25	30.22
2	A2	−00−	25	45	60	20	27.90
3	C1	−0−0	25	45	40	25	25.28
4	B1	−0+0	25	45	80	25	35.09
5	A3	−00+	25	45	60	30	29.92
6	A4	−+00	25	70	60	25	29.49
7	D1	0−0−	50	20	60	20	27.55
8	C2	0—0	50	20	40	25	24.72
9	B2	0−+0	50	20	80	25	30.75
10	E1	0−0+	50	20	60	30	24.83
11	C3	00−−	50	45	40	20	27.10
12	B3	00+−	50	45	80	20	29.49
13	D2	0000	50	45	60	25	27.02
14	D3	0000	50	45	60	25	30.36
15	E2	0000	50	45	60	25	29.68
16	G1	00−+	50	45	40	30	26.18
17	F1	00++	50	45	80	30	37.35
18	D4	0+0−	50	70	60	20	26.34
19	C4	0+−0	50	70	40	25	26.11
20	B4	0++0	50	70	80	25	31.36
21	E3	0+0+	50	70	60	30	26.69
22	H1	+−00	75	20	60	25	13.08
23	E4	+00−	75	45	60	20	16.27
24	G2	+0−0	75	45	40	25	15.45
25	F2	+0+0	75	45	80	25	17.75
26	H2	+00+	75	45	60	30	13.62
27	H3	++00	75	70	60	25	16.51

. The relationship between TFC and the independent variables was modelled using the second-order regression equation as follows: Y = 29.02 − 7.10X₁ + 0.4458X2 + 3.08X₃ + 0.3276X₄ + 1.04X₁X₂ − 1.88X₁X₃ − 1.17X₁X₄ − 0.1930X₂X₃ + 0.7678X₂X₄ + 2.20X₃X₄ − 6.06X₁² − 1.43X₂² + 0.9421X₃² − 0.7301X₄².

The ANOVA results for the second-order regression model (

Table 4. ANOVA results of the experimental mode.

Sources	Sum of Mean Squares	Degrees of Freedom	Mean of Square	F-Value	p-Value
Model	1019.27	14	72.81	30.43	<0.0001
X1	604.93	1	604.93	252.85	<0.0001
X2	2.38	1	2.38	0.9967	0.3378
X3	113.76	1	113.76	47.55	<0.0001
X4	1.29	1	1.29	0.5382	0.4773
X1X2	4.33	1	4.33	1.81	0.2036
X1X3	14.09	1	14.09	5.89	0.0319
X1X4	5.46	1	5.46	2.28	0.1567
X2X3	0.1489	1	0.1489	0.0623	0.8072
X2X4	2.36	1	2.36	0.9857	0.3404
X3X4	19.28	1	19.28	8.06	0.0149
X12	196.17	1	196.17	82.00	<0.0001
X22	10.88	1	10.88	4.55	0.0543
X32	4.73	1	4.73	1.98	0.1849
X42	2.84	1	2.84	1.19	0.2971
Residues	28.71	12	2.39
Model fit error	22.47	10	2.25	0.7203	0.7063
Random error	6.24	2	3.12
Sum	1047.98	26
R2: 0.9726, Adjusted R2: 0.9406, Predicted R2: 0.8631
Suitability: 19.9602, CV (%): 6.00

) indicate a good fit with the experimental data. The R² value is 0.9726, showing that 97.26% of the variation in total flavonoid content was explained by the model. The predicted R² (0.8631) and adjusted R² (0.9406) were both above 0.8, with a difference of less than 0.2, confirming strong predictive reliability. Adequate precision reached 19.9602 (>4), indicating a high signal-to-noise ratio and model suitability for optimization. The lack-of-fit test yielded a p-value of 0.7063 (>0.05), suggesting no significant deviation between the model and actual data. The coefficient of variation (CV) was 6.00%, below the 10% threshold, confirming the model’s stability and repeatability. Regression coefficients, including the intercept, linear, quadratic, and interaction terms, were estimated based on the second-order polynomial model. ANOVA results (Table 4) show that parameters with p-values < 0.05 were statistically significant. In this study, X₁ (solvent concentration), X₃ (temperature), interactions X₁X₃ and X₃X₄, and the quadratic term X₁² significantly influenced total flavonoid content. In contrast, other parameters with p-values > 0.05 were not statistically significant and could be excluded from the regression equation. The refined regression equation is as follows: Y = 29.16 − 7.10X₁ + 0.4458X₂ + 3.08X₃ + 0.3276X₄ − 1.88X₁X₃ + 2.20X₃X₄ − 6.12X₁² − 1.48X₂².

Response Surface Optimization Analysis of Flavonoid Extraction Conditions

The influence of all independent variables on the TFC was analyzed in the second-order regression model by the response surface plot (Figure 3) to find out the optimal conditions for ultrasonic-assisted flavonoid extraction from Physalis angulata. Flavonoids are predominantly present in a glycosidic form with a low Log P; thus, the polarity of extracting solvents affects their extraction [12]. Indeed, according to the single-variable tests, the polarity, demonstrated by its dilution with water, is the most influential factor on flavonoid yield (p < 0.0001). As shown in Figure 3a–c, TFC increased with the ethanol concentration, reaching its maximum at approximately 40%. Beyond this level, a marked decrease in extraction yield was observed when the ethanol concentration exceeded 55%. This trend can be attributed to changes in the polarity of the solvent system; at higher ethanol concentrations, the reduced polarity likely limits the solubilization and diffusion of polar flavonoid compounds from the plant matrix. The optimal solvent polarity for flavonoid extraction was found at around 50% ethanol, consistent with previous reports indicating maximum polyphenol recovery within the 40–60% ethanol range. Ultrasonic extraction time (20–70 min) exhibited no statistically significant effect on TFC (p = 0.3378), as indicated in Figure 3a,d,e. Flavonoid yield remained relatively stable across the tested time range, suggesting that equilibrium may have been achieved early in the process. This implies that extraction time within the investigated interval can be adjusted flexibly without substantially affecting the recovery of flavonoids.

Extraction temperature demonstrated a strong positive effect on TFC (p < 0.0001), as presented in Figure 3b,d,f. Increasing the temperature from 60 to 80 °C significantly enhanced flavonoid yield. The improved extraction performance at elevated temperatures can be attributed to increased molecular mobility, enhanced diffusion rates, and more efficient mass transfer between the solvent and plant material. These results confirm temperature as a key variable in ultrasonic-assisted extraction optimization.

The solvent-to-material ratio alone did not exert a statistically significant influence on TFC (p = 0.4773), as shown in Figure 3c,e. However, its interaction with temperature (Figure 3f) revealed a significant synergistic effect (p = 0.0149). Specifically, the extraction yields increased as the ratio rose from 25:1 to 30:1 (mL/g), indicating that this parameter may enhance extraction efficiency under higher temperature conditions

Overall, these results demonstrate that ethanol concentration and extraction temperature are the primary factors governing flavonoid extraction efficiency from Physalis angulata under ultrasonic conditions, whereas ultrasonic time and solvent-to-material ratio play secondary or interaction-dependent roles. Most flavonoids can be extracted in 40 min [12], and in our work, up to 49 min for the optimal efficiency. Prolonged sonication increases energy consumption and degradation risk without significant yield improvement. Temperature had no significant standalone effect, but its interaction with solvent-to-material ratio was notable. High temperature can enhance diffusion and solubility, but in combination with low solvent volume, it may cause localized overheating and flavonoid degradation. Flavonoid degradation products have been identified as quercetin-3-O-β-D-glucopyranoside (isoquercitrin) and other polar compounds [34]. The dissolution of these polar compounds in a less polar solvent is less efficient due to the polarity of the subcritical. As the temperature rises, the water level declines. Sharifi (2013) reported that flavonoid degradation occurred in subcritical water extraction at above 110 °C for 10 min [36].

These reports emphasize the importance of multivariable optimization. The Box–Behnken design and RSM, the model achieved a strong fit (R² = 0.9726). Predicted optimal conditions, including 31.66% ethanol, 48.73 min, 80 °C, 30:1 mL/g, yielded a TFC of 38.09 ± 1.70 mg RU/g (

Table 5. Experimental validation results of the optimized parameters (n = 3).

Optimal Values				TFC (mg RU/g)
Ethanol Concentration (%)	Sonication Time (min)	Extraction Temperature (°C)	Solvent-to-Material Ratio (g/mL)	Predicted	Experimental
31.66%	48.73	80 °C	1:30	38.09 ± 1.70 *	34.58 ± 0.87 *

* No significantly statistical difference (p > 0.05).

). Experimental validation (n = 3) yielded 34.58 ± 0.87 mg RU/g. Statistical analysis showed no significant difference between the predicted and experimental values (p > 0.05), confirming the reliability of the regression model. The experimental result was reproducible and fell within the predicted range, demonstrating that the model is suitable and effective for simulating and optimizing flavonoid extraction from Physalis angulata.

The strong correlation between experimental and predicted values validates the RSM approach, enabling a detailed understanding of variable interactions. For instance, temperature and solvent ratio can synergize or conflict, affecting solubility and stability. Multivariate analysis is essential to avoid suboptimal or degrading conditions. The Box–Behnken design also proved its efficiency by capturing linear, quadratic, and interactive effects with minimal experiments, which is critical in phytochemical research for the materials with high variability. Compared to traditional methods, modern experimental designs like RSM offer high reproducibility and predictive power for process development. These results align with earlier studies using ultrasonic-assisted extraction, reinforcing the method’s feasibility and adaptability to other botanical sources. For example, Giordano et al. [37] extracted flavonoids from kiwi peel using ultrasonication. The experimental conditions were optimized using SRM, and the results showed that UAE was significantly affected by ultrasonication time and ethanol concentration. In a report of Zhang et al. (2021), ethanol concentration and temperature were key variables affecting flavonoid extraction from Prunella vulgaris [38]. As reported by Soares et al. (2022) about Azadirachta indica, extraction conditions were optimized at 30% ethanol, 0.2 g/mL solid–liquid ratio, and 33 min sonication [39].

3.3. Development and Application of UV-Vis Spectroscopy to Determine Total Flavonoid Content

The AlCl₃ colourimetric method was introduced by Christ and Müller in 1960 [28] but has been improved many times to increase its sensitivity and selectivity. Our improved version consists of NaNO₂–AlCl₃–NaOH [40,41], the observed sequential colour changes correspond to chemical interactions of the formation and stabilization of the flavonoid–Al³⁺ complex. Firstly, NaNO₂ selectively reacts with ortho-dihydroxy groups (catechol) on the B-ring of flavonoids to form a stable pale-yellow nitroso or o-quinone derivatives that alter the molecule’s electron distribution. The subsequent addition of AlCl₃ makes the yellow colour darker as Al³⁺ ions chelate with 3-OH/4-keto or 5-OH/4-keto positions, expanding the π-conjugated system and shifting light absorption to longer wavelengths. Finally, NaOH creates an alkaline environment for deprotonating phenolic OH groups to form phenolate ions, which bind more strongly to Al³⁺ and the nitroso complexes to intensify the colour of the complexes for spectroscopic measurement [41,42,43,44]. This modification increases sensitivity, shifts the absorption maximum to a longer region (a bathochromic shift) instead of 415–430 nm in the original method, and therefore reduces interference such as proteins, pigments, and phenolics from matrix compounds that do not react specifically with Al³⁺ [42,44,45].

In the validation, the complexon product created from 40 µg/mL rutin standard was scanned in the 450–800 nm range to choose a wavelength of 510 nm as the maximum wavelength for flavonoid quantification. At this wavelength, there was a positive and strong correlation between the absorbance and the flavonoids concentration (R² = 0.9996 ≥ 0.998), the result of evaluating the significance of the regression coefficients gave us the final regression equation as follows y = 0.0099x + 0.0661 (Significance F = $6.1168\times {10}^{-8}$ < α = 0.05) in the flavonoids concentration range of 15–120 µg/mL. LOD was 2.48 µg/mL. The calculated LOQ from the regression equation was 7.52 µg/mL; nevertheless, since concentrations below 15 µg/mL showed high noise and poor stability, the validated LLOQ was established at 15 µg/mL, which also defined the working quantitation range.

The UV-Vis spectrophotometric method was validated at 510 nm according to AOAC guidelines; all the parameters met all the requirements, as follows:

-

Intra-day precision: The RSD value of all TFC-determined samples was 1.89%, less than the allowable limit of 3.7% (for the concentration of analytes of 0.1%).

-

Inter-day precision: the same samples were analyzed on two different days, and the average of RSD was 2.21%, below 3.7%. F-Test Two-Sample for Variances and t-Test Two-Sample Assuming Equal Variances were used to test the variance and mean, respectively, of the two-day measurement data. The testing shows no significant difference, which indicates that the method met the precision requirements.

-

Accuracy: recoveries were 99.52 ± 2.07%, 102.28 ± 1.78%, 104.06 ± 0.79% for the concentration levels of 80%, 100% and 120%, respectively. At each level, the recovery rates were in the range of 95–105% and the RSD values were all less than 3.7%, which confirms that the method met the accuracy requirements.

The optimized extraction and validated UV-Vis method were applied to determine the TFC in leaves collected in Nam Đinh (located in the North of Vietnam), Tien Giang (located in Mekong Delta), Can Gio mangrove biosphere reserve and An Giang located in the South of Vietnam. To have a relative comparison, the plants were at the same growth stage (green mature with fruits) and collected in the morning of the dry season to minimize seasonal variation. Since the samples were processed by the same procedure, the differences in TFC can be attributed to the environmental factors such as sunlight exposure, soil type, and local microclimate. Environmental stressors, such as drought or salinity, can induce the biosynthesis of secondary metabolites, including flavonoids, as part of the plant’s defense response [46]. Sunlight exposure, particularly UV-B radiation, is known to enhance flavonoid accumulation by activating phenylpropanoid pathway enzymes [47]. Soil conditions influence nutrient availability, which can also affect metabolite profiles, particularly, nitrogen limitation has been linked to higher phenolic content in several plant species [48]. Indeed, as seen in Figure 4, Tien Giang, belonging to Mekong delta with fertile soil and hot climate were found to accumulate the highest flavonoids (91.89 ± 0.34 mg/g), compared to other locations such as the mangrove forest in Can Gio with cool weather and less sunlight exposure.

As far as the authors have searched in the literature, only one paper was published in 2021 about the total flavonoids in Vietnam, but in different locations (Kien Giang, Ca Mau, Đong Thap) [49]. Kien Giang and An Giang are 100 km apart. Similarly, Đong Thap and Tien Giang are 100 km apart. A comparative assessment indicated that the TFCs in the present study were relatively higher. This enhancement is likely due to the use of a more efficient extraction solvent, which facilitated improved the solubilization and recovery of flavonoid compounds. In comparison with the study of Indonesia, water gave a very low total flavonoid extraction efficiency [50], compared to the organic solvents. A study of P. angulata in Nigeria also used ethanol, but with the addition of acetic acid, reported the TFC in the range of our study (18.7 mg/g) [51]. These differences highlight the importance of standardization or uniformity of research methods in scientific works. In order to directly compare flavonoid content, these procedures need to be standardized and optimized in factors such as extraction method, and analytical techniques.

The validated spectrophotometric method demonstrated reliable quantification of total flavonoid content for a rapid comparison of samples collected from different regions. However, potential interferences from non-flavonoid components such as proteins, free amino acids, sugars, organic acids, tannins, and various pigments can interfere with the UV-Vis determination of flavonoids. Such components may absorb light within similar spectral regions or participate in side reactions, leading to inaccuracies in quantification. These substances contribute to background absorbance, induce baseline instability through light scattering by colloidal particles, or exhibit overlapping absorption bands with flavonoids, typically within the 250–450 nm range, thereby compromising analytical specificity. To mitigate this effect, aluminium chloride complexation was employed to induce a bathochromic shift in the absorption spectrum of flavonoids, enhancing spectral selectivity [42,45]. A wavelength of 510 nm was selected to be specific for flavonoid–aluminium complexes such as those of rutin, quercetin, and luteolin, as well as certain phenolic acids [42].

Centrifugation and filtration were applied to eliminate particulate matter responsible for light scattering, especially chlorophyll, the most obviously important interference. As a result, the sedimentation technique used for chlorophyll removal also contributed to the reduction in endogenous interferences [27,28]. The TFC strongly depends on solvent polarity, temperature, partly on solvent or solid ratio and sonication time. Highly polar solvents tend to co-extract phenolic acids, whereas low-polarity solvents dissolve the aglycone moieties of flavonoid glycosides. Prolonged or high-energy ultrasonic treatment may induce free-radical formation, leading to partial degradation of target compounds. Therefore, the experimental conditions were optimized to achieve an appropriate balance of solvent polarity suitable for most flavonoid glycosides while avoiding heating above 100 °C to minimize thermal degradation. The procedure maintained a controlled sequence and duration of sonication, ensured stable room temperature during the colour development phase, and required synchronous measurement of samples and standards to reduce potential signal drift over time. Although some polyphenols may also react with aluminium chloride, the employed NaNO₂–AlCl₃ colourimetric system under alkaline environment has been widely accepted for TFC in natural samples. To correct background matrix effects, a blank extract and spiked standards were utilized. Furthermore, spectral preprocessing, including baseline correction, and chemometric deconvolution approaches such as variable-wavelength selection through multivariate calibration are highly recommended to enhance analyte resolution in complex matrices [52].

Despite the limited selectivity and tolerance for minor matrix-induced deviations, the UV-Vis spectrophotometric approach remains highly advantageous due to its simplicity, low operational cost, and rapid throughput. Therefore, it represents an efficient and practical tool for preliminary assessment of flavonoid content before more sophisticated chromatographic analyses. For a more detailed understanding of the flavonoid composition in Physalis angulata L. extracts, future investigations could integrate this green UV-Vis workflow with chromatographic techniques. Such a combined approach would enhance analytical specificity, reduce matrix interferences, and facilitate precise quantification of individual flavonoids, thereby contributing to a deeper characterization of the plant’s phytochemical diversity and supporting comprehensive pharmacological evaluations.

3.4. A Comprehensive Greenness Assessment of the Workflow of Optimized Extraction and Quantification for Total Flavonoid Content

The complete analytical workflow for green score calculation is briefly described: an exact weight of 2 g of dried leaf powder was extracted with the RSM-optimized conditions: 60 mL of 32% ethanol, ratio of solvent to material of 30:1 (mL:g), ultrasonication in 49 min at 80 °C. After removal of residual ethanol, the extract was washed with distilled water (1:1), followed by a dechlorophyllization by sedimentation at 4 °C in 24 h. The mixture was then centrifuged at 10,000× g for 30 min at 4 °C to collect the supernatant for UV-Vis spectroscopic assay. A total of 1 mL was transferred to a 10 mL volumetric flask, followed by the addition of 4 mL of distilled water and 0.3 mL of 5% NaNO₂ solution. After 5 min, 0.3 mL of 10% AlCl₃ solution was added to the reaction in 6 min. Finally, 2 mL of 1 M NaOH was added, and the solution was filled up to the volume with distilled water, which was then incubated at room temperature for 10 min, and its absorbance was recorded at 510 nm.

Although this is considered a “classic” method, it is still included in many pharmacopoeial and AOAC protocols for herbal material analysis and widely recommended in recent studies as a reliable operational measure to compare TFC between extraction conditions [38,39,50,53], due to its simplicity, rapidity, low cost, and applicability to a wide range of plant samples. UV-Vis spectroscopic determination scored a higher green score compared to HPLC, according to different greenness evaluation tools: lower energy use (<0.1 kWh per sample) as Eco-Scale score [19], simple instrumentation available in most labs as BAGI score [23], aligning with Principle 9 “the use of energy should be minimised” and receiving a full 1.0 score (analytical systems consume <0.1 kWh per sample, ultrasound-assisted extraction) as AGREE score [22]. This method is easily calibrated with standards such as quercetin or rutin that are available and cheaper than gallic acid (for calibration with Folin–Ciocalteu reagent). This makes the whole workflow more eco-efficient and so increases the green score.

Eco-Scale, firstly proposed for green organic synthesis assessment, was then adapted to evaluate analytical methods. The total score for the greenest workflow is 100 on the scale; the higher score means that the process is greener and more economical. Our complete workflow scored 86 points in total (

Table 6. A greenness evaluation profile of the complete procedure from sampling to the results (Eco-Scale, AGREE, GAPI, BAGI).

Eco-Scale Score		AGREE
Reagent	Penalty Points
Ethanol	4
Sodium hydroxide	2
∑	6
Instrument	Penalty points
UV-Vis	0
Occupational hazards	0
Wates (25 mL)	5
No treatment	3
∑	8
Total penalty points	14
Eco-Scale score	86
ComplexMo GAPI		BAGI

), higher than 75 points, which means excellent green analysis [19].

GAPI method uses five coloured pentagrams to evaluate the environmental impacts of the analytical procedure in different steps, from sampling, method, sample preparation, solvent/reagent use, and energy consumption [20]. A complex modified GAPI (ComplexGAPI) generates a cumulative score along with a pentagram [21]. This scoring system considers the range of choices within each category, including sample preparation, reagents/solvents, instrumentation and sample preservation, transport and storage from the top left to the bottom left of the pentagrams. The total points are summed and divided by the maximum achievable points to determine the percentage score (76% in Table 6).

The Analytical GREEnness calculator (AGREE), a comprehensive, flexible, and straightforward assessment, is based on 12 principles of green analytical chemistry (SIGNIFICANCE). The final score is transformed into a pictogram from a 0 to 1 scale; the assessment of 12 segments is reflected by the colour, in which the dark green colour represents the highest score [22]. The overall score shown in the middle of the pictogram of our work is 0.61. In Table 6, our greenest points are principles 6 (derivatization avoided) and 9 (minimized the use of energy). The least green factor is principle 10 with red colour because our reagents were not from a renewable source.

The final green evaluation method is BAGI, focusing on the practice of White Analytical Chemistry [23]. The blue pictogram is constituted by 10 attributes that represent quantitative analysis, multi-element analysis, simple instrumentation available, sample preparation, samples processed per hour, reagents and materials, preconcentration, semi-automation, and amount of sample. The dark blue, blue, light blue, and white colours score 10, 7.5, 5, and 2.5 points, respectively, which are aggregated to a total of 62.5 points (Table 6), demonstrating its applicability.

4. Conclusions

The procedure for quantifying total flavonoids by UV-Vis spectroscopy with rutin as a standard at 510 nm wavelength was validated on the standard and actual samples according to AOAC guidelines (2023) for specificity, linearity in the validated range of 15–120 µg/mL (R² = 0.9996), and precision (recovery rates from 99.52 to 104.06%). The LOD was 2.48 µg/mL. The calculated LOQ (10σ/S) was 7.52 µg/mL, but the validated lower limit of quantification (LLOQ) was set at 15 µg/mL in the working quantitation range to avoid the noise baseline. The sedimentation method was found to be effective in removing chlorophyll pigments and was suitable for application in optimization modelling studies. Ultrasonic extraction is a green and innovative extraction process to receive the high flavonoid content in a short time. The extraction conditions affecting the yield of flavonoids were investigated and optimized using response surface modelling with Box–Behnken design. The optimal extraction conditions include 31.66% ethanol, a sonication time of 48.73 min at 80 °C and a solvent/material ratio of 30:1 (mL/g). This resulted in a predicted total flavonoid content of 38.09 mg RU/g. The statistical analysis shows no significant difference compared to the measured value from the experiment (p > 0.05), which confirms the suitability and accuracy when applying the optimization model in practice. The success of this research has miniaturized the extraction of total flavonoids from Physalis angulata with the support of RSM method, which contributes to establishing a scientific framework for standardizing this plant species, in turn, supports the creation of high-value natural therapeutic products. This aligns with the growing tendency for modernized medicine integration with traditional practices in the country. Further studies on this plant are recommended to isolate, identify, characterize and elucidate the structures of the potential bioactive compounds.