Food Chemistry: X 22 (2024) 101484

Contents lists available at ScienceDirect

Food Chemistry: X
journal homepage: www.sciencedirect.com/journal/food-chemistry-x

Optimization of ultrasonic assisted ethanolic extraction for natural

pigments from butterfly pea flower applied in Thai dessert using
Box-Behnken approach
Samart Sai-Ut a, *, Apisara Teksee a, Jaksuma Pongsetkul b, Sirima Sinthusamran c,
Saroat Rawdkuen d
a
Department of Food Science, Faculty of Science, Burapha University, Chonburi, Thailand
b
School of Animal Technology and Innovation, Institute of Agricultural Technology, Suranaree University of Technology, Nakhon Ratchasima, Thailand
c
Department of Agricultural Education, School of Industrial Education and Technology, King Mongkut’s Institute of Technology Ladkrabang, Bangkok, Thailand
d
Food Science and Technology Program, School of Agro-Industry, Mae Fah Luang University, Chiang Rai, Thailand

A R T I C L E I N F O A B S T R A C T

Keywords: Butterfly pea is a natural color source used in food and dessert. This study optimized ultrasound-assisted
Butterfly pea flower extraction with ethanol for pigments from butterfly pea flowers (BPF) using a Box-Behnken method. Key fac­
Colorant tors explored were solid-to-solvent ratio, ultrasound extraction time, and ethanol concentration. The extracted
Response surface methodology
compounds were evaluated for extraction yield (EY), total phenolic content (TPC), total anthocyanin content
Ultrasound
(TAC), and DPPH antioxidant activity. EY increased significantly with reduced ethanol concentration. Optimal
conditions were predicted and experimentally validated. BPF extracts showed distinct absorption wavelengths at
different pH levels. BPF extract was used in coconut milk jelly, resulting in the lowest b* value. These findings
highlight the value of optimal ultrasonic-assisted extraction for enhancing BPF’s natural colorant extraction and
promoting sustainable use in food and dessert applications.

1. Introduction Butterfly pea powder, derived from the blue pea flower, is a well-known
product. BPF contains various bioactive compounds, including antho­
Color plays a significant role in the acceptability and attractiveness cyanins, flavonoids, glycosides, steroids, resins, and phenols. These
of food, often influencing consumers’ perceptions of taste and quality. compounds contribute to its anti-diabetic, antioxidant, antibacterial,
Consumer acceptance and perception of food quality are linked to color anti-inflammatory, and analgesic properties (Chusak et al., 2019).
attributes. The rising consumer demand for natural ingredients and Acylated anthocyanins in C. ternatea are known for their structural
concerns related to nutrition and health have driven the adoption of complexity and stability, which makes them attractive for use as natural
natural color sources in the food and beverage sector. Natural colors in colorants in the food and beverage industry (Giusti & Wrolstad, 2003;
plants come from various pigments: red and yellow hues come from Jeyaraj, Lim, & Choo, 2022). The acylation provides structural stability
betalains, green from chlorophyll, red and purple from anthocyanins, and impacts their color properties, making them more resistant to
and yellow and orange from carotenoids (Ghosh, Sarkar, Das, & Chak­ degradation by light, heat, and pH changes compared to non-acylated
raborty, 2022). Natural pigments originating from edible fruits, vege­ anthocyanins (Jokioja, Yang, & Linderborg, 2021). The traditional
tables, and flowers are not only perceived as safe and clean but are also Thai dessert known as butterfly pea coconut milk jelly, renowned for its
acknowledged for their potential health benefits when utilized as food historical significance and widespread popularity, incorporates vibrant
ingredients (Ghosh, Sarkar, Das, & Chakraborty, 2021). colors derived from butterfly pea flowers. The increasing demand for
Butterfly pea flowers (BPF), known as botanic name Clitoria ternatea, utilizing this natural extract in industrial-scale production underscores
part of the Fabaceae family, are a natural source of food coloring. BPF the necessity for an extraction process that is not only suitable but also
serve culinary and medicinal purposes. Their extract is used as a natural aligns with the requirements of both producers and consumers.
food colorant, and their roots and leaves are used in herbal drinks. Extraction of plant pigments using solvents is well-reported and

* Corresponding author at: Department of Food Science, Faculty of Science, Burapha University, Chonburi 20131, Thailand.
E-mail address: [email protected] (S. Sai-Ut).

https://doi.org/10.1016/j.fochx.2024.101484
Received 4 December 2023; Received in revised form 15 May 2024; Accepted 15 May 2024
Available online 17 May 2024
2590-1575/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
S. Sai-Ut et al. Food Chemistry: X 22 (2024) 101484

simple, but it has disadvantages such as long exposure times and lower three levels per factor to accommodate a second-order regression model,
yields. Consequently, there is significant interest in “green” extraction resulting in a reduced number of experiments compared to the CCD,
technologies, which offer shorter processing times, higher yields, and which requires five levels per factor. Therefore, the objective of this
lower environmental risks, addressing the limitations of conventional study was to assess the physicochemical and color properties of butterfly
methods like Soxhlet and heat-assisted extraction. Ultrasonic assisted pea and employ RSM for the optimization of UAE parameters (ultrasonic
extraction (UAE) is a promising technology for efficiently retrieving time, solid-to-solvent ratio, and ethanol concentration) for determining
natural pigments from by-products (Linares & Rojas, 2022). This the yield, TPC, TAC and antioxidant activity of butterfly pea flower
approach involves the utilization of sound waves with specific proper­ extract using the BBD.
ties, including frequency, amplitude, and wavelength, to induce physi­
cochemical alterations in the medium. Ultrasonic transducers produce 2. Materials and methods
these waves by converting electrical energy into sound waves charac­
terized by their desired frequency and intensity (Kentish, 2017). UAE 2.1. Materials preparation
offers multiple advantages, such as reduced extraction and processing
duration, lower energy consumption, decreased carbon dioxide emis­ BPFs were sourced from the local market in Bang Sean, Chonburi,
sions, and minimized solvent usage. These benefits result in enhanced Thailand, in July 2023, at coordinates latitude of 13.2688◦ N and
extraction yields and the potential for employing cleaner and environ­ longitude of 100.9274◦ E. BPFs were hand-harvested after approxi­
mentally friendly extraction processes, particularly with water or sol­ mately 45 days of growth, preferably in the early morning, before noon.
vents (Chemat et al., 2017). Moreover, UAE optimizes the extraction of BPFs were promptly subjected to drying in a drying oven (Model:
heat-sensitive components that might otherwise yield unsatisfactory UM500, Memmert, Germany) at 50 ◦ C for 18 h, resulting in a moisture
results. However, UAE also has limitations. It can require significant content reduction to around 15% (wet basis). The desiccated BPF sample
energy, especially at higher power, leading to higher costs. Prolonged were subsequently gathered and transformed into a powdered form
use generates heat, potentially degrading heat-sensitive compounds like utilizing a mixer grinder at speed of 32,000 rpm (Model: YB-800 A, Yun
anthocyanins, thus necessitating temperature control. The initial in­ Bang, China). Subsequently, the BPF powder underwent sieving with a
vestment in ultrasonic equipment is high, and its operation requires laboratory sieve, with all particles passing through an 80-mesh sieve
specialized knowledge (Kumar, Srivastav, & Sharanagat, 2020). Scaling being collected in a polypropylene plastic bag for subsequent analysis.
up to industrial scale is challenging due to variations in ultrasound
transmission and cavitation in larger volumes. Intense ultrasonic power
2.2. BPF extraction by UAE
can break down bioactive compounds, reducing their efficacy and
requiring careful optimization of extraction parameters.
The UAE process utilized an ultrasonic bath (Tru-sweep 950D, Crest,
For the efficient optimization of extraction parameters for bioactive
USA) with a 2.5 L capacity, operating at 250 watts of power and a fre­
compounds from plants, response surface methods (RSM) are a valuable
quency of 45 kHz. BPF powder was prepared into a beaker size 100 mL
tool. In this context, RSM has been employed in a limited number of
and used for each experiment according to the Box-Behnken experi­
studies involving UAE, often utilizing a range of solvents like water,
mental design (Tables 1), which follows a three-level, three-factor full
methanol, ethanol, and others for the extraction process. While certain
factorial design with center points. The three independent variables
solvents may offer improved extraction efficiency, they can also pose
included the sample-to-solvent ratio: g:mL (X1), ultrasonic extraction
health risks and toxicity concerns. Considering factors such as cost,
time: time (X2), and ethanol concentration: % (X3). Throughout the
health implications, and safety, water stands out as the most suitable
extraction procedure, the initial temperature was set at room tempera­
solvent. RSM strategies offer advantages in terms of time and labor ef­
ture (28 ◦ C). To prevent the temperature from surpassing 35 ◦ C,
ficiency compared to other approaches (Chakraborty, Uppaluri, & Das,
meticulous control was exercised by modulating the water level within
2020). This is primarily due to the reduced number of experimental
the ultrasonic bath. This adjustment ensured that the temperature did
trials required to analyze numerous parameters and their interactions.
not exceed 35 ◦ C. Upon completion of the extraction process, the sample
Many studies have employed widely used RSM designs, such as the
solution was centrifuged at 8000 g for 15 min using a Hermle centrifuges
central composite design (CCD) and the Box-Behnken Design (BBD). The
(HERMLE Labortechnik, Germany) and then the supernatant was
BBD, primarily used in most RSM investigations, is specifically devel­
filtered through Whatman filter paper (no. 4). The filtered samples were
oped for fitting second-order models and is notable for requiring only
stored in amber flasks at 4 ◦ C, sealed, and capped for subsequent

Table 1
BBD of three variables with their observed responses of BPF extracts.
Run* Independent variables EY TPC TAC DPPH
(%) (mg GAE/ g dried extract) (mg/100 g dried extract) (μmole TE/mg dried extract)
X1 X2 X3
(g:mL) (min) (%)

1 20 30 60 41.8 ± 0.72 7.81 ± 0.08 55.91 ± 2.48 87.18 ± 3.42

2 40 30 60 44.55 ± 0.49 13.51 ± 0.23 56.27 ± 2.68 113.59 ± 7.51
3 20 60 60 45.47 ± 0.23 7.23 ± 0.28 42.32 ± 0.47 82.55 ± 5.66
4 40 60 60 48.89 ± 0.11 10.81 ± 0.24 51.73 ± 2.13 106.2 ± 3.30
5 20 45 40 46.32 ± 1.71 7.17 ± 1.26 51.65 ± 2.25 92.36 ± 8.51
6 40 45 40 50.2 ± 0.45 9.52 ± 0.72 50 ± 2.32 107.33 ± 4.34
7 20 45 80 37.71 ± 0.69 7.48 ± 0.33 39 ± 1.13 93.32 ± 5.99
8 40 45 80 41.13 ± 0.65 13.9 ± 0.58 39.95 ± 0.98 131.81 ± 1.21
9 30 30 40 44.83 ± 0.94 8.43 ± 0.89 51.53 ± 2.36 109.11 ± 4.38
10 30 60 40 46.7 ± 0.78 8.16 ± 0.44 44.56 ± 1.81 114.54 ± 2.28
11 30 30 80 41.44 ± 3.35 13.12 ± 3.94 35.73 ± 0.86 112.52 ± 7.69
12 30 60 80 40.55 ± 1.71 9.19 ± 1.2 38.77 ± 0.88 124.67 ± 6.13
13 30 45 60 43.39 ± 0.94 9.52 ± 0.36 54.52 ± 2.92 111.59 ± 1.81
14 30 45 60 42.67 ± 0.95 9.31 ± 0.8 53.28 ± 2.16 128.6 ± 2.01
15 30 45 60 42.04 ± 0.74 9.11 ± 0.71 51.78 ± 1.39 118.31 ± 2.35
*
Randomized experiments were conducted.

2
S. Sai-Ut et al. Food Chemistry: X 22 (2024) 101484

analyses. where ΔAbsorbance is the absorbance differences for each sample by

subtracting the absorbance at the higher pH from the absorbance at the
2.3. Experimental design and optimization lower pH. Molar absorptivity is a constant for anthocyanins at the given
wavelength. ε is molar extinction coefficient of cyanidin-3-glucoside
In this study, a BBD with three levels and three factors was employed (26,900 M− 1 cm− 1). l represents the cell’s path length, typically set at
to assess the impact of process variables related to UAE on response 1 cm.
variables. Three specific extraction process variables, denoted as X1
(solid-to-liquid ratio, ranging from 1:20 to 1:40 (g:mL)), X2 (ultrasonic 2.6. Analysis of total phenolic compounds (TPC)
extraction time, between 30 and 60 min), and X3 (ethanol concentration,
varying from 40% to 80%), were selected. A total of 15 experimental The analysis of TPC was determined following the Folin-Ciocalteu
runs were conducted randomly as outlined in Table 1. The data collected method (Swain & Hillis, 1959). 50 μL of the extract, 200 μL of de-
during these experiments were used to develop a comprehensive second- ionized water, and 50 μL of Folin-Ciocalteu reagent were combined in
order polynomial model, yielding the regression coefficients presented a quartz vial and thoroughly mixed using a Vortex. After allowing the
in Eq. (1). mixture to react for 5 min, 500 μL of a 7% (w/v) Na2CO3 solution was
∑4 ∑4 3
∑ 4
∑ added and mixed well. The solution was then incubated at room tem­
Y = β0 + β Xi + β X2 + βij Xi Xj (1) perature in the dark for 60 min. Subsequently, the absorbance was
i=1 i i=1 ii i
i=1 j=i+1 measured at 760 nm using a UV-spectrophotometer (10S UV–Vis Spec­
In this equation, Y represents the response function, where β0 is the trophotometer, Genesys, USA), and the results were expressed in gallic
intercept, and βi, βii, and βij are the coefficients for the linear, quadratic, acid equivalents (mg GAE/100 g dry sample) utilizing a gallic acid
and interactive terms, respectively. The variables Xi, X2i , and XiXj (0–50 mg/mL) standard curve. In cases where the absorbance value
represent the coded independent variables, respectively. exceeded the linear range of the standard curve, additional dilution was
The analysis of variance (ANOVA) was performed using Minitab performed.
version 18 software to determine the significance of interactions within
the model (p < 0.05). Statistical significance was confirmed by an F-test. 2.7. Analysis of antioxidant activity using the DPPH method
The fitness of the polynomial model equation was verified by analyzing
the lack of fit using an F-test. Results with p < 0.05 were considered The assessment of antioxidant activity was performed by employing
significant, and they were checked against the lack of fit. Coefficient of the DPPH method, with modifications following Brand-Williams et al.
determination R2, adjusted R2, and adequate precision were also (1995). A volume of 500 μL of BPE extracts, diluted one hundred times
examined for model suitability. Graphical and numerical optimization with distilled water, is combined with an equivalent volume of 500 μL of
techniques were employed to determine optimal independent variable 0.5 mM DPPH solution (in 95% ethanol). The reaction solution was
levels. Confirmation experiments were then conducted to validate the adequately mixed with a Vortex machine and kept in the dark for 30
optimal conditions. min. Subsequently, absorbance was measured at a wavelength of 517
nm using a UV-spectrophotometer. A standard curve was established
2.4. Analysis of extraction yield (EY) using Trolox, and the results are reported in Trolox equivalent antioxi­
dant capacity (μmole TE/ mg dried extract).
To determine the extraction yield, the solvent was removed by
evaporation, and the remaining sample was considered as the yield. 2.8. Impact of pH on the color change of BPF extract
Briefly, 50 mL of the BPF extract were transferred to a stainless-steel
moisture container and subsequently placed into a hot air oven The color response of the BPF extract to pH changes in aqueous
(UM500, Memmert, Germany) set at 105 ◦ C for 18 h. After the drying media (pH 2–12) was determined. pH buffer solutions (concentration of
process, the dried samples were weight and calculated using the Eq. (2). 0.1 M) at pH 2, pH 4, pH 6, pH 8, pH 10, and pH 12 were prepared using
HCl/KCl buffer, KHP (potassium hydrogen phthalate) /HCl buffer,
Extraction yield = (weight obtained after extraction/initial weight)
KH₂PO₄/NaOH buffer, Glycine/NaOH buffer, and KCl/NaOH buffer,
x 100 respectively. BPE was diluted one hundred times with distilled water
(2) and dispensed into test tubes. Buffer solutions at different pH were
added into the test tubes to observe the color change in response to
2.5. Analysis of total anthocyanin content (TAC) variations pH. Spectra of BPF extract at different pH were measured
using the absorbance values utilizing a spectrophotometer in the
The total anthocyanin content is determined as follows, with modi­ wavelength range of 300 to 700 nm.
fications based on Lee, Durst, and Wrolstad (2005). BPF extract was
diluted to an appropriate concentration using distilled water. Each 2.9. Color changes of BPF extract during storage
sample was divided into two set, one being mixed with 0.2 M KCl, pH
1.0, and the other with 0.2 M acetate buffer (pH 4.5). Subsequently, In this study, samples obtained under optimized conditions were
these mixtures were incubated at room temperature for a duration of 5 concentrated using a rotary evaporator (R-100, BUCHI, Switzerland) at
min to facilitate equilibrium attainment. The absorbance measurements 45 ◦ C and a pressure of 120 millibars. The obtained BPF extract (with the
were then performed at 510 and 700 nm using disposable cuvettes with concentration of 20 mg dried extract/mL) stored in amber bottles was
a 1 cm path length on a UV–Vis spectrophotometer (10S UV–Vis Spec­ incubated at 35 ◦ C for 0, 1, and 2 weeks. Subsequently, the BPE extracts
trophotometer, Genesys, USA), while instrumental zero calibration was were subjected to color measurement using a Hunter Lab Colorimeter
achieved using distilled water. The ΔAbsorbance values were employed (MiniScan XE Plus, HunterLab, USA). The colorimeter equipment was
to estimate the total anthocyanin concentration in the sample through calibrated using standard white and black tiles or other reference sam­
the application of Beer’s Law. The equation is as follows Eq. (3): ples. The BPF extract solution was transferred into the sample holder
Total anthocyanin = ΔAbsorbance × Molar absorptivity cup, which was then placed within the measurement area, covering the
(3) entire measurement surface. The recorded color values are expressed in
× Dilution factor/ε × l
the CIE LAB system including L* (lightness), a* (red-green), and b* value
(yellow-blue).

3
S. Sai-Ut et al. Food Chemistry: X 22 (2024) 101484

2.10. Impact of BPF extract concentration in coconut milk jelly on the governing the UAE process.
color properties The ANOVA suggested that quadratic, second-order polynomial
models were the most suitable choice for effectively describing the re­
The investigation focused on the impact of varying concentrations of sponses during the experimental procedures in the UAE process. The
obtained BPF extract in coconut milk jelly (a Thai dessert) on color derivation of predictive quadratic models involved the use of a back­
appearance. In summary, the preparation involved combining 45 g of ward elimination regression approach. Statistically non-significant
agar powder with 300 mL of water, stirring thoroughly, and letting it terms were eliminated, with the retention of those deemed essential to
rest for 10–15 min before heating. Coconut water (300 mL) was then maintain the hierarchical structure. The factors ranged from − 1 to +1,
added, and the mixture was heated over medium heat with constant and their impact on the response variables was evaluated using corre­
stirring until the agar powder dissolved. Sugar (280 mL) and salt (5 g) lation coefficients (R2) with statistical significance at p < 0.05. The
were incorporated, ensuring complete dissolution. After bringing the relationship between the factors and responses was considered signifi­
mixture to a boil, the heat was reduced to low, and coconut milk was cant (p < 0.05) based on the quadratic polynomial model. Furthermore,
added while stirring continuously for 2 min. BPF extract was then added the lack of fit test yielded statistically non-significant results (p > 0.05),
at different concentrations (0.5%, 1%, 2%, and 5%) for each treatment. as indicated by the calculated F values for the EY, TPC, TAC, and DPPH
The ultimate pH of the mixed solution was 6.5. The solution was activity, respectively. Additionally, small pure error values were
transferred into flower-shaped molds, left to set in the refrigerator (4 ◦ C) observed, signifying the data’s good repeatability and a satisfactory
for 1 h, and then demolded for subsequent color analysis using a Hunter coefficient of determination. The analysis of the relationship between
Lab Colorimeter. the studied factors and the response values revealed a regression equa­
tion model with significance at p < 0.05. Table 2 displays the effects of
all the examined parameters, their corresponding F-values, and the
2.11. Statistical analysis
associated probability (p-value). These coefficients were computed
using the coded values of the parameters. The correlation coefficient of
The experiment was conducted in triplicate. Data and variances were
the EY reached 0.8997, while the adjusted R2 value was 0.7192. This
analyzed using ANOVA, and means were compared using Tukey’s test at
suggests that the model can predict the EY closely in accordance with the
a 95% confidence level. Correlation analysis was performed using the
actual experimental results (p < 0.05). The analysis revealed that X1 and
Pearson correlation method. All analyses were conducted using Minitab
X3 had a significant linear influence on the EY (p < 0.05) as indicated by
18 statistical software.
F-values of 7.30 (p -value = 0.043). Conversely, X2 did not exert a sig­
nificant linear influence on the EY (p > 0.05). The results were fitted
3. Results and discussion
using multiple regression, and the relationship between the EY of BPF
and the variables was expressed by the following Eq. (4):
3.1. Regression and variance analysis (ANOVA)
EY = 57.66–0.684X1 –0.224X2 + 0.0749X3 + 0.0142X1 2 (4)
The RSM technique, utilizing a Box-Behnken design, was employed
In the analysis of TAC, the model exhibited a relative value of 0.9375
to conduct a 3-factor, 3-level experiment for the optimization of color­
and an adjusted R2 value of 0.8249 (Table 3). It was observed that X2
ants derived from butterfly pea flowers through UAE process. A
had a significant linear influence on TAC (p < 0.05) with an F-value of
comprehensive set of 15 experimental designs is detailed in Table 1. A
7.11 and a P -value of 0.045. The linear effect of ethanol concentration
regression model was utilized to establish the relationship between the
(X3) on TAC was significant, as well as the quadratic effect of ultrasound
independent variables (extraction parameters) and the dependent vari­
extraction time on TPC. When considering the squared term relation­
ables (EY, TPC, DPPH, and TAC). The results of the experiment
ship, it was evident that X3 significantly affected TAC (p < 0.05) as a
demonstrated a range of outcomes: the total yield ranged from 37.71 to
single factor. Nevertheless, when examining the interaction effect of two
50.20 mg/100 g, while the total phenolic compound content varied
factors (2-way interaction), no significant interaction was observed be­
between 7.17 and 13.90 mg GAE/g sample. Additionally, the anthocy­
tween the factors (p > 0.05). Consequently, an equation to predict the
anin values spanned from 35.73 to 56.27 mg/100 g sample, and the
TAC of BPFs was derived by the following Eq. (5):
antioxidant activity, assessed by the DPPH method, ranged from 82.55
to 131.81 μmol TE/mg sample, as detailed in Table 1. This compre­ TAC = 1.9–0.1838X2 + 2.231X3 + 0.0043X1 2 –0.0209X3 2 (5)
hensive analysis provides valuable insights into the various parameters

Table 2
Variance analysis for the fitted models (EY and TPC).
Source DF EY TPC

Adj SS Adj MS F p-value Adj SS Adj MS F p-value

Model 9 139.368 15.4854 4.98 0.046* 70.4406 7.8267 110.24 0.000*

X1 1 22.680 22.6801 7.30 0.043* 40.7253 40.7253 573.61 0.000*
X2 1 10.103 10.1025 3.25 0.131 6.9938 6.9938 98.51 0.000*
X3 1 92.616 92.6161 29.81 0.003* 13.5460 13.5460 190.79 0.000*
X21 1 7.965 7.9651 2.56 0.170 0.0940 0.0940 1.32 0.302
X22 1 3.757 3.7572 1.21 0.322 0.4975 0.4975 7.01 0.046*
X23 1 0.399 0.3991 0.13 0.735 0.0073 0.0073 0.10 0.761
X1X2 1 0.112 0.1122 0.04 0.857 1.1236 1.1236 15.83 0.011*
X1X3 1 0.053 0.0529 0.02 0.901 4.1412 4.1412 58.33 0.001*
X2X3 1 1.904 1.9044 0.61 0.469 3.3489 3.3489 47.17 0.001*
Error 5 15.532 3.1065 0.3550 0.0710
Lack-of-fit 3 14.620 4.8732 10.68 0.087 0.2709 0.0903 2.15 0.333
Pure error 2 0.913 0.4563 0.0841 0.0420
Total 14 154.901 70.7956
R2 89.97% 99.50%
R2(adj) 71.92% 98.60%

DF: degrees of freedom; SS: sum of squares; MS: mean square; * stand for significant differences (p < 0.05).

4
S. Sai-Ut et al. Food Chemistry: X 22 (2024) 101484

Table 3
Variance analysis for the fitted models (TAC and DPPH).
Source DF TAC DPPH

Adj SS Adj MS F p-value Adj SS Adj MS F p-value

Model 9 641.673 71.297 8.33 0.016* 2741.07 304.56 5.53 0.037*

X1 1 10.260 10.260 1.20 0.324 1339.55 1339.55 24.31 0.004*
X2 1 60.830 60.830 7.11 0.045* 3.86 3.86 0.07 0.802
X3 1 245.311 245.311 28.65 0.003* 189.93 189.93 3.45 0.122
X21 1 0.693 0.693 0.08 0.787 894.25 894.25 16.23 0.010*
X22 1 15.847 15.847 1.85 0.232 158.77 158.77 2.88 0.150
X23 1 265.307 265.307 30.99 0.003* 18.98 18.98 0.34 0.583
X1X2 1 20.430 20.430 2.39 0.183 1.90 1.90 0.03 0.860
X1X3 1 1.690 1.690 0.20 0.675 138.30 138.30 2.51 0.174
X2X3 1 25.100 25.100 2.93 0.148 11.29 11.29 0.20 0.670
Error 5 42.805 8.561 275.46 55.09
Lack-of-fit 3 39.026 13.009 6.88 0.129 128.67 42.89 0.58 0.681
Pure error 2 3.779 1.890 146.79 73.40
Total 14 684.478 3016.53
R2 93.75% 90.87%
R2(adj) 82.49% 74.43%

DF: degrees of freedom; SS: sum of squares; MS: mean square; * stand for significant differences (p < 0.05).

(A) (B) 60
Yield.
< 42
42 – 43
55 43 – 44
44 – 45
48 45 – 46
50 46 – 47
> 47

46 Time
Yield (% ) 45

44
60 40

42 50
Time (min ) 35
40
20
30 30
40 30
Ratio (g:mL ) 20 25 30 35 40
Ratio

(C) 80
Yield. (D) 80
Yield.
< 40 < 40
40 – 42 40 – 42
42 – 44 42 – 44
44 – 46 44 – 46
70 70
46 – 48 > 46
> 48
EtOH

EtOH

60 60

50 50

40 40
20 25 30 35 40 30 35 40 45 50 55 60
Ratio Time

Fig. 1. Response surface plots of EY as affected: (A) time and ratio; contour plot between (B) time and ratio, (C) ethanol concentration and solid-to-liquid ratio, and
(D) time and ethanol concentration, respectively.

The linear and interaction terms of X1, X2, and X3 were significant (p
< 0.05) for TPC. The TPC demonstrated exceptionally high R2 and TPC = 0.89 + 0.0794X1 + 0.0857X2 + 0.0497X3
adjusted R2 values, indicating a strong correlation with the model (R2 = + 0.001567X2 2 –0.003533 X1 X2 + 0.005087 X1 X3− 0.00305 X2 X3
0.995, adjusted R2 = 0.986). In contrast, only the linear and quadratic (6)
terms of X1 were significant (p < 0.05) for DPPH. The relationships
between the responses (TPC and DPPH) and variables were interpreted DPPH = − 83.9 + 8.97X1 + 2.74X2 –0.638X3 –0.1574X1 2 –0.0299X2 2
through the subsequent second-order polynomial Eq. (6) and (7), + 0.0294 X1 X3
respectively. (7)

5
S. Sai-Ut et al. Food Chemistry: X 22 (2024) 101484

Consistent with our findings, Koraqi, Petkoska, Khalid, Kumar, and respective p-values of X1 (0.043) and X3 (0.003), indicating an interac­
Pareek (2023) also observed an increase in TPC as the ultrasound- tion effect between variables X1 and X3. This finding was consistent with
assisted extraction time increased, suggesting a positive quadratic previous research by Baskaran, Mudalib, and Izirwan (2019), which
relationship. The solvent volume was identified as the second most reported that the extraction yield of anthocyanins from BPFs increased
influential factor positively affecting TPC after the extraction time. In a with greater solvent volume. Fig. 1(D) reveals that X2 (extraction time)
study on ultrasound-assisted extraction from Saussurea lappa, longer with a longer duration and X3 with a lower concentration result in
extraction times were associated with a negative effect on antioxidant increased yield. This outcome was due to the beneficial effects of high-
capacity, as indicated by a coefficient of − 8.74, suggesting that extended frequency sound waves in extracting phytochemicals, which were bio­
extraction times may not enhance, and could potentially reduce, anti­ logically active compounds. These sound waves facilitated mass move­
oxidant capacity (Ali & Ahmed, 2023). ment and improved solvent penetration into the plant materials used for
extraction, thereby enhancing extraction efficiency. In the case of plant
materials, the application of high-frequency sound waves promoted
3.2. The effect of various UAE conditions on the EY swelling, breaking cell walls, and releasing substances in greater quan­
tities. Normally, BPF extracts, rich in anthocyanins and carotenoids,
The effect of various components on the EY is illustrated in Fig. 1. It influence food acceptability and properties. Anthocyanins, responsible
was evident that X1 (sample-to-solvent ratio) and X2 (extraction time) for the flower’s blue color, stabilize color and offer antioxidants (Vidana
exhibit an increasing trend in yield when the ratio was 40 (w/v) and the Gamage, Lim, & Choo, 2021). BPF extracts, abundant in anthocyanins
extraction time was 60 min. This observation aligns with Fig. 1(B), and carotenoids, impact food qualities. Anthocyanins stabilize color and
which demonstrates that higher X1 (1:40) and a longer extraction time of provide antioxidants, while carotenoids contribute yellow-orange hues
60 min significantly increased the EY (p < 0.05). This effect can be and antioxidant benefits. Additionally, C. ternatea contains various
attributed to the increase in the temperature in the solution. During the bioactive compounds like alkaloids, tannins, glycosides, and flavonoids,
UAE process, the application of ultrasonic waves generates cavitation alongside essential minerals such as calcium, magnesium, potassium,
bubbles in the liquid medium. The rapid formation and collapse of these and zinc, as well as fatty acids like palmitic acid, stearic acid, petrose­
bubbles produce localized high temperatures and pressures, leading to linic acid, and linoleic acid (Multisona, Shirodkar, Arnold, & Gramza
an overall increase in the temperature of the solution. This elevated Michalowska, 2023). Moreover, BPFs contained polysaccharides that
temperature enhances the mass transfer and solubilization of the target were more soluble in water than in ethanol, potentially leading to the
compounds, thereby improving the extraction efficiency. Fig. 1(C) in­
extraction of non-pigment substances and increasing the overall product
dicates that a higher concentration of X1 and a lower concentration of X3 weight (Soria & Villamiel, 2010).
had a significant influence on the yield. This was supported by the

Fig. 2. Response surface plots of TAC as affected: (A) time and ratio; contour plot between (B) time and ratio, (C) ethanol concentration and solid-to-liquid ratio, and
(D) time and ethanol concentration, respectively.

6
S. Sai-Ut et al. Food Chemistry: X 22 (2024) 101484

3.3. The effect of various UAE conditions on the TAC thus decreasing the extraction efficiency. Therefore, the highest
extraction rate was achieved within a specific ethanol concentration
In Fig. 2(A), as ethanol concentration increases, anthocyanin content range where the solvent properties are most favorable for anthocyanin
decreases, suggesting an inverse relationship, consistent with Fig. 2(D) solubilization. Goula, Ververi, Adamopoulou, and Kaderides (2017)
where extraction time (X2) should not exceed 40 min. The interaction found that solid-to-liquid ratio of 100 g/L was optimal for anthocyanin
effect between the variables is further demonstrated in Fig. 2(B), which extraction from wine lees, with a slight decrease as the ratio decreased to
presents a contour plot illustrating the influence of X1 and X2. It was 25 g/L. More and Arya (2021) as well as Chen, Yang, Mou, and Kong
evident that higher anthocyanin content was achieved with a combi­ (2018) studied this factor within the ranges of 20–100 g/L and
nation of a short extraction time and a greater solid-to-solvent ratio. 28.6–66.7 g/L, respectively. The consensus from these studies is that
Fig. 2(C) displays a contour plot highlighting the influence of X1. Ac­ increasing the solid-to-liquid ratio decreases the extraction yield,
cording to the findings of Zhang, Fan, Khan, Yan, and Beta (2020), aligning with mass transfer phenomena expectations, where a higher
extremely short or excessively long extraction times did not yield high gradient enhances the driving force for mass transfer. However, lower
levels of anthocyanin. In cases of too short an UAE time, anthocyanins solid-to-liquid ratios may lead to increased solvent consumption,
might not fully dissolve in the extraction solvent. Conversely, extended necessitating investigation for each specific case to determine the opti­
extraction at high temperatures could lead to the degradation of an­ mum ratio. The impact of ultrasound power intensity on anthocyanin
thocyanins, as these compounds were temperature sensitive. Addition­ extraction is noteworthy; for instance, increasing ultrasonic power up to
ally, as reported by Li, Zhang, Wang, Feng and Han (2023), ethanol was 300 W maximized anthocyanin content from mulberry wine residues,
more effective in extracting anthocyanins than deionized water. The but further increases reduced yield (Zhang et al., 2020). Similar trends
relationship between ethanol concentration and the extraction rate of were observed in anthocyanin extraction from jabuticaba peel, where
anthocyanins was not linear. According to Fig. 2, the extraction rate of increasing ultrasound intensity up to 7.3 W/cm2 increased anthocya­
anthocyanins was highest within a certain range of ethanol concentra­ nins, but at 13 W/cm2, the yield decreased (Gadioli Tarone et al., 2021).
tions, rather than continuously increasing with higher ethanol concen­ However, power level variations may not consistently affect pigment
trations. This phenomenon can be attributed to the dual role of ethanol extraction, as observed in the ultrasonic-assisted extraction of antho­
as both a solvent and an extraction enhancer. At optimal ethanol con­ cyanins from grape pomace (da Rocha & Noreña, 2020). Concerning
centrations, the solvent polarity is balanced, which effectively solubi­ processing time, samples with higher concentration may require longer
lizes anthocyanins and maximizes their extraction. However, beyond extraction times for effective cell disruption and increased pigment
this optimal range, further increases in ethanol concentration may release, as demonstrated in anthocyanin extraction from eggplant peel
reduce the solvent’s ability to interact with water-soluble anthocyanins, cut into squares (Ferarsa et al., 2018). Conversely, excessively long

(A) (B) 60
TPC.
< 8
8 – 9
55 9 – 10
10 – 11
11– 12
50 12 – 13
> 13
14
Time

12 45
TPC (mg GAE/g)
10
40
80
8
60 EtOH (% ) 35
20
30 40
40 30
Ratio (g:mL ) 20 25 30 35 40
Ratio
(C) 80
(D)
80
TPC. TPC.
< 8 < 9
8 – 10 9 – 10
10 – 12 10 – 11
12 – 14 11 – 12
70 > 14 70 > 12
EtOH

EtOH

60 60

50 50

40 40
20 25 30 35 40 30 35 40 45 50 55 60
Ratio Time

Fig. 3. Response surface plots of TPC as affected: (A) time and ratio; contour plot between (B) time and ratio, (C) ethanol concentration and solid-to-liquid ratio, and
(D) time and ethanol concentration, respectively.

7
S. Sai-Ut et al. Food Chemistry: X 22 (2024) 101484

extraction times can lead to reduced yield (Zafra-Rojas et al., 2020). The stages of extraction, solutes on the surface of the plant material were
discussion emphasizes the importance of ultrasonic power and ampli­ dissolved, which could be further enhanced with increased extraction
tude for the extraction of compounds from plants. However, in this time. Subsequently, a slower extraction process was initiated, involving
experiment, the ultrasonic power and amplitude were kept constant and a gradual release of bioactive compounds from the plant matrix into the
were not included as independent variables in the Box-Behnken design. solvent. In the RSM plot of Fig. 3(C), the TPC exhibited the highest value,
While ultrasonic power and amplitude are indeed critical factors influ­ reaching 14 mg GAE/g sample at the condition of ethanol concentration
encing extraction efficiency, the focus of this study was to optimize the of 80% and the ratio of 1:40 (g:mL). This is consistent with other studies
conditions related to solvent concentration, extraction time, and that demonstrated the significant impact of ethanol concentration on
material-to-solvent ratio. Future studies could incorporate ultrasonic phenolic compound extraction. For instance, citrus limon fruit extracted
power and amplitude as variables to further refine and optimize the with 63.93% ethanol yielded 1502.2 ± 0.88 mg GAE/100 g (Dahmoune
extraction process. et al., 2013), while Pistacia lentiscus leaves obtained a lower TPC value of
142.76 ± 19.98 mg GAE/g when using 46% ethanol for extraction.
(Dahmoune et al., 2014). Additionally, Esmaeelian, Jahani, Feizy, and
3.4. The effect of various UAE conditions on the TPC
Einafshar (2020) reported a lower TPC value from saffron corms extract
(100.39 mg GAE/100 g dry saffron corm) using 80% ethanol compared
Fig. 3 illustrates the influence of extraction conditions on TPC.
to our study (900–1300 mg GAE/100 g). The results of this experiment
Higher solid-to-solvent ratio correlated with increased phenolic content,
demonstrated that the use of ultrasonic waves aided in achieving good
supported by Fig. 3(C) where high solid-to-solvent ratio combined with
extraction efficiency. Ultrasonic waves generate microjets, causing dis­
elevated ethanol concentration results in greater phenolic content. Fig. 3
ruptions in the cell walls, a phenomenon known as cavitation, which
(B) presents a contour plot revealing the influence of X1 and X2
accelerated the mass transfer rate and facilitates the release of plant
(extraction time). It indicated that a high X1 and a low X2 led to higher
compounds into the solvent. Arruda et al. (2019) investigated the impact
phenolic compound content. Conversely, Fig. 3(D), a contour plot
of high-intensity ultrasound parameters, including nominal power
depicting the influence of X2 and X3, suggests that a low X2 (short
(160–640 W) and extraction time (0.5–5.0 min), on the recovery of
extraction time) and a high X3 (ethanol concentration) resulted in
phenolic compounds from araticum peel. The study revealed a signifi­
increased phenolic compound content. These observations aligned with
cant influence of these process parameters on phenolic recovery, indi­
the findings of Leong, Juliano, and Knoerzer (2017), who noted that
cating that ultrasound-assisted extraction disrupts plant cell walls
increasing the amplitude enhances the efficiency of breaking cell walls,
through cavitation phenomena.
facilitating the release of important compounds into the solvent. How­
ever, an extended extraction time might lead to decreased extraction
efficiency due to two distinct mechanisms. Initially, during the early

(A) (B) 60
DPPH.
< 90
90 – 100
55 100 – 110
110 – 120
> 120
50

120
Time

110 45
DPPH umoleTE/mg
100
40
60
90
50
Time 35
40
20
30 30
Ratio 40 30
20 25 30 35 40
Ratio
(C) 80
(D)
DPPH.
< 100
80
DPPH.
100 – 110
110 – 120 < 110.0
110.0– 112.5
120 – 130
70 112.5– 115.0
> 130
115.0– 117.5
70
117.5– 120.0
120.0 – 122.5
122.5 – 125.0
EtOH

60 > 125.0
EtOH

60

50

50

40
20 25 30 35 40
Ratio 40
30 35 40 45 50 55 60
Time

Fig. 4. Response surface plots of DPPH as affected: (A) time and ratio; contour plot between (B) time and ratio, (C) ethanol concentration and solid-to-liquid ratio,
and (D) time and ethanol concentration, respectively.

8
S. Sai-Ut et al. Food Chemistry: X 22 (2024) 101484

3.5. The effect of various UAE conditions on the DPPH optimized condition predicts the following values: EY = 45.29%, TAC =
54.09 mg/100 g sample, TPC = 12.76 mg GAE/g sample, and DPPH =
The impact of extraction conditions on antioxidant activity, assessed 115.5 μmole TE/mg sample. In the pursuit of finding the ideal condi­
through the DPPH method, is depicted in Fig. 4. With the increase of the tions, all three variables were taken into consideration. The combined
solid-liquid ratio, the DPPH scavenging ability initially rises and then analysis of all four variables yielded desirability values of 0.6688 (EY),
decreases. This indicates that there was an optimal solid-liquid ratio 0.89427 (TAC), 0.8307 (TPC), and 0.60693 (DPPH), resulting in an
where the DPPH scavenging activity was maximized, rather than a linear overall suitability value for the model of 0.7411, with a typical range of
increase with higher solid-liquid ratios. This behavior suggests that 0 to 1, where a value closer to 1 indicates a more appropriate model. Our
beyond a certain point, the extraction efficiency may be hindered by too study achieved optimal UAE conditions for extracting phenolics and
much solid material, leading to decreased DPPH scavenging ability. The antioxidants from BPF that showed similar results from the different
contour plot in Fig. 4(B) suggests that peak antioxidant activity was plant materials. Melgar et al. (2019) who optimized the extraction of
achieved with higher values of both solid-to-solvent ratio and time, bioactive compounds from Opuntia engelmannii peel using extraction
approximately within the range of 45–50 min. Fig. 4(C) presents a time (t), solid-to-liquid ratio (S/L), methanol concentration, and tem­
contour plot illustrating that heightened X1 and increased concentra­ perature (T) reported that the optimum values were achieved at t = 2.5
tions of X3 (ethanol concentration) contribute to elevated antioxidant min, S/L = 5 g/L, metOH = 34.6%, and T = 30 ◦ C, resulting in maximum
activity. Additionally, Fig. 4(D) indicates that a shorter extraction time responses of 13.9 mg of phenolic acids/g, 2.4 mg of flavonoids/g, and
(low X2) coupled with a high X3 concentration led to increase the anti­ 71.8% of extractable solids. Umair et al. (2021) determined that the
oxidant activity. These findings are in line with the research of Berkani optimal conditions for ultrasound-assisted extraction of carotenoids
et al. (2020), which reported different solvent concentrations resulting from carrot pomace were 17 min of extraction time, a temperature of
in various polarities led to varying solubilities of the substance. Water 32 ◦ C, and an ethanol concentration of 51%. Additionally, our study’s
serves as an efficient solvent for extracting polar molecules, while for optimized conditions featured higher ethanol concentration and longer
less polar compounds, greater efficiency can be achieved with ethanol. extraction time compared to the conditions used by Plazzotta, Ibarz,
An increase in ethanol concentration significantly reduced the solvent’s Manzocco, and Martín-Belloso (2020) in their work on antioxidant
polarity. The decrease in solvent polarity with increasing ethanol con­ compounds from peach waste.
centration led to reduced extraction of substances from the plant ma­ The model verification utilized selected cases based on the optimi­
terial. This was because less polar solvents may not efficiently extract zation results obtained from the response surface model. The confir­
certain compounds compared to more polar solvents. Additionally, the mation experiments conducted under the optimal conditions predicted
high antioxidant activity of the extracted anthocyanins could be by the model to assess the model’s accuracy, a comparison was made
attributed to their hydroxyl-rich structure, particularly on the B-ring, between the actual experimental value and the predicted value. Imple­
which enables effective free radical scavenging (Kongpichitchoke, Hsu, menting the optimized conditions for colorant extraction from BPFs
& Huang, 2015). This structural configuration was responsible for the resulted in the EY of 41.01 ± 3.74 g/100 g sample, TAC of 48.49 ± 2.27
heightened antioxidant effects observed in materials rich in anthocya­ mg/100 g sample, TPC of 13.68 ± 1.85 mg GAE/g sample, and DPPH of
nins and phenolic compounds. It had been documented in previous 129.53 ± 0.161 μmole TE/mg sample. Comparison of the experimental
studies that such compounds exhibit elevated antioxidant potential due value and the predicted value demonstrated similar results, validating
to their high number of hydroxyl groups (Benavente-Garcı́a, Castillo, the accuracy of the model. These experimental values closely matched
Lorente, Ortuño, & Del Rio, 2000; Kongpichitchoke et al., 2015). The the predicted values. Furthermore, when considering percent error, it
inhibition of DPPH was linked to the abundance of phenolic compounds was determined to be 9.45, 10.35, 7.21, and 12.15 for EY, TAC, TPC, and
(Berkani et al., 2020). Additionally, other compounds like ascorbic acid, DPPH, respectively. Within an acceptable threshold of 10%, validate the
tocopherol, and pigments may contribute synergistically to overall reliability of the conditions for colorant extraction from BPFs predicted
antioxidant activity, indicating a role beyond phenolic compounds alone by the mathematical model, affirming its utility for predictive purposes.
(Benhanifia, Mohamed, Bellik, & Benbarek, 2013). Ultrasound extrac­
tion proved more efficient in obtaining antioxidant-rich polyphenols 3.7. Impact of pH on the color change of BPF extract
from BPF. The enhanced recovery and antioxidant activity were attrib­
uted to ultrasound’s mechanical cavitation. The study investigated the impact of pH on the color change of BPE
extract. The results, as displayed in Fig. 5, revealed significant variations
3.6. Optimization and validation in the extract’s color based on the pH levels. At pH 2, the extract showed
maximum absorption at 300 nm, absorbing green light and producing a
Utilizing the response surface model, we employed numerical opti­ pink color. At pH 4, the highest absorption occurred at 575 nm, indi­
mization techniques to determine the optimal UAE conditions for col­ cating strong absorption of yellow light and a purple hue. At pH 6, the
orants from BPF. The objective was to identify the combination of λmax was 625 nm, signifying absorption of orange light and a blue color.
variables that would yield the highest predicted value in all response Between pH 8 and 12, the extract displayed an λmax at 300 nm,
(EY, TAC, TPC, and DPPH). The optimal conditions for independent absorbing red light and resulting in a green color. In various pH con­
variables and the predicted values of the response were determined as ditions, anthocyanins display a range of colors from red in acidic envi­
follows: a solid-to-solvent ratio of 1:40 (g/mL), an ethanol concentration ronments to purple in neutral pH and blue at higher pH levels. The
of 37% (v/v), and an extraction time of 61 min. As shown in Table 4, the stability of red-colored anthocyanins, existing as flavylium cations, is
higher in lower pH solutions. The solubility of anthocyanins increases in
Table 4 lower pH due to the formation of highly soluble flavylium cations.
Predicted and experimental values under the optimal operating conditions. However, as pH rises, the increased deprotonation rate of flavylium
Response Unit Predicted Experimental cations results in reduced color stability. Anthocyanin-tannin polymer­
variable value value* ization enhances color stability in lower pH conditions. At higher pH,
EY % 45.29 41.01 ± 3.74
colorless carbinol pseudobase and chalcone structures form, leading to
TAC mg/100 g sample 54.09 48.49 ± 2.27 the generation of anionic quinonoidal species (Fossen, Cabrita, &
TPC mg GAE/g sample 12.76 13.68 ± 1.85 Andersen, 1998). This occurs due to the kinetic and thermodynamic
DPPH μmole TE/mg 115.5 129.53 ± 1.61 competition between the hydration reaction of flavylium ions. The blue
sample
quinonoidal species is unstable at lower pH, and at pH 4–5, an antho­
*
The value represents the mean of three replications with standard error. cyanin solution exhibits minimal hue. At neutral pH, resonance-

9
S. Sai-Ut et al. Food Chemistry: X 22 (2024) 101484

Fig. 5. Relationship between the wavelength and the absorbance of the colorant from butterfly pea flowers at pH 2–12.

stabilized quinonoid anions, representing the purple color of anthocy­ 3.8. Color changes of BPF extract during storage
anins, form through further deprotonation of the quinonoidal species
(Coutinho, Freitas, & Maçanita, 2015). The alteration in the color of BPE The color variations observed in the pigment extracted from BPF
extracts across various pH levels is attributed to changes in the structure under optimal UAE conditions, with a concentration of 20 mg dried
of anthocyanin pigments, which depend on the number of hydroxyl extract/mL, during storage at 35 ◦ C for 0, 1, and 2 weeks exhibited
groups within the molecule (Hasanah, Mohamad Azman, Rozzamri, minor alterations, as illustrated in Fig. 6. The brightness value (L*)
Zainal Abedin, & Ismail-Fitry, 2023). Anthocyanins are classified as slightly increased in the range of 0.51–0.6, indicating a tendency toward
phenolic compounds, characterized by a chromophore with a benzene enhanced brightness compared to the initial conditions at week 0. The
ring. The groups attached to the chromophore affect the molecule’s positive a* value, which represents redness, increased within the range
absorption spectrum, shifting it toward longer wavelengths (Campbell, of 3.1–3.64 when compared to week 0. Conversely, the b* value, signi­
Pearson, & Marble, 2019). These findings align with previous research fying yellow color, was positive but displayed a decrease within the
that indicated the stability of anthocyanin depends on factors such as range of 0.68–0.71. These results are consistent with research by
pH, with low pH levels enhancing stability, as well as the influence of Charurungsipong, Tangduangdee, Amornraksa, Asavasanti, and Lin
light (Hasanah et al., 2023). This phenomenon, known as Bathochromic (2020), who found that a decrease in anthocyanin content with pro­
or a red shift, refers to the alteration in the absorption or emission longed heating time, and concurrently, higher temperatures were asso­
wavelengths of a substance, often resulting in a change in color. This ciated with lower anthocyanin content. The anthocyanin stability
shift can occur when various environmental factors, including pH levels depends on several factors, including the structure of the molecule, the
and the type of solvent, change. It is an essential characteristic of certain presence and number of sugar molecules near the flavylium ion, the
compounds, including anthocyanins, which are known for their sensi­ number of glycosidic acid linkages, and the number and position of
tivity to changes in pH and solvent conditions (Wiyantoko, 2020). The substitutions on the flavylium ion (Khoo, Azlan, Tang, & Lim, 2017). For
pH sensitivity of BPF anthocyanins could prove advantageous for their instance, 3-deoxy anthocyanins, which have a yellow color due to
use in food colorants. Anthocyanins can adjust their color based on pH, dihydroxylation of the third carbon, are more stable than red-colored 3-
making them adaptable to various food products with different acidity hydroxy anthocyanins. The addition of more hydroxyl groups to the
levels. This versatility allows for a broad spectrum of colorful options phenyl ring will darken the color, while the replacement of hydroxyl
without relying on synthetic additives. Moreover, the natural vibrancy groups with methoxyl groups at the 3′ and 5′ positions enhances the red
and stability of anthocyanin colors enhance the visual appeal of food, hue (Amić, Davidović-Amić, & Trinajstić, 1993). Anthocyanins exhibit
meeting consumer preferences for natural and clean-label ingredients. reduced stability at elevated solution temperatures. A study of Mori,
Thus, BPF anthocyanins offer versatility, visual appeal, and alignment Goto-Yamamoto, Kitayama, and Hashizume (2007) found that heat
with consumer demands, making them ideal for natural food coloring in treatment up to 35 ◦ C significantly decreased total anthocyanin content
diverse applications. in common grapes compared to control berries at 25 ◦ C. However, heat
treatment of anthocyanin-rich extracts may not lead to pigment degra­
dation, as these extracts typically contain phenolic compounds that
inhibit enzymatic degradation by polyphenol oxidase. Moreover, mild
heat treatment, such as blanching, in the food processing industry has
been demonstrated to prevent anthocyanin oxidation by inactivating the
enzymatic reaction up to 50 ◦ C (Patras, Brunton, & O’Donnell, 2010).
The primary anthocyanins found in C. ternatea are ternatins, which are
characterized by their acylation and glycosylation patterns. Acyl groups
in these anthocyanins form intramolecular bonds that enhance struc­
tural stability and result in deeper, more stable colors. They stabilize the
chromophore, making the pigments less susceptible to color loss under
varying pH levels and light exposure (Bakowska-Barczak, 2005). Addi­
tionally, acyl groups act as barriers against enzymatic degradation and
oxidation, which commonly affect non-acylated anthocyanins
(Houghton, Appelhagen, & Martin, 2021). This finding provides valu­
Fig. 6. Color changes of BPF extract during storage at 35 ◦ C for 2 weeks. able insights into the stability of natural colorants derived from butterfly

10
S. Sai-Ut et al. Food Chemistry: X 22 (2024) 101484

Table 5
Color properties and appearance of the coconut milk jelly product incorporating colorants derived from BPF, ranging from 0 to 5% (w/w) concentration.
Color value* Concentration (%)

0 0.5 1 2 5

L* 44.68 ± 1.32a 37.32 ± 1.10b 27.49 ± 1.60c 21.29 ± 1.56d 17.53 ± 1.81d
a* − 1.00 ± 1.47a − 4.28 ± 1.54a − 9.61 ± 1.56b − 7.71 ± 1.73b − 4.64 ± 0.95a
b* − 3.52 ± 0.68a − 4.66 ± 1.27a − 13.66 ± 1.42bc − 11.64 ± 1.64b − 16.62 ± 1.50c
Appearance

Different letters denote significant differences (p < 0.05) between concentrations.

*
The value represents the mean of three replications with standard error.

pea extract, offering understanding of colorant changes during storage. traditional rice ball had high sensory value and low constant rate of
Such information is crucial for implementing these colorants in the food anthocyanin degradation when adding 10% BPF extract. It also reported
and beverage industries, especially considering the factors of storage that heat and processing lead to anthocyanin loss in the foods. Higher
time and conditions. extract utilization ratios are associated with better anthocyanin reten­
tion under the same heating time. Steaming processes, as observed in
black rice, can cause an 88% reduction in anthocyanin content due to
3.9. Impact of BPF extract concentration in coconut milk jelly on the color
thermal degradation (Surh & Koh, 2014). This loss influences the color
properties
change observed in cakes containing butterfly pea flower extract after
steaming. Thermal processing in the food industry can degrade antho­
The different concentration (0, 0.5, 1, 2, and 5% w/w) of BPF extract
cyanins, and researchers have proposed various pretreatments to in­
as food colorant in coconut milk jelly products was investigated. In this
crease extraction yield and enhance color pigment stability
experiment, the sample achieved a final pH of 6.5. It was observed that
(Charurungsipong et al., 2020). Temperature was found to influence
as the concentration increased, the coconut milk jelly product became
anthocyanin stability, with higher temperatures accelerating anthocy­
darker, resulting in reduced brightness (lower L* value). Additionally,
anin degradation. Olivas-Aguirre et al. (2016) reported that anthocyanin
the jelly product displayed a deeper green hue (lower a* value) with the
decomposition rates double for every 10 ◦ C increase in storage tem­
lowest a* value occurring at a 1% concentration, leading to a bluer color
perature. It has been reported that cyanidin 3-glucoside and cyanidin 3-
(increased b* value). Table 5 illustrates that the jelly product exhibits a
rutinoside decompose at 100 ◦ C in weakly acidic solutions (pH 1–4),
deeper blue color with an increase in BPF extract concentration. Given
both in the presence and absence of oxygen. Anthocyanins may further
the importance of color appearance for consumer acceptance in dessert,
degrade into coumarin derivatives through decomposition or polymer­
it becomes crucial to investigate the suitability of adding colorants to
ization and may form complex compounds (Li, Zhang, Wang, Feng, &
food products. Anthocyanin pigments are a valuable natural food
Han, 2023). Chemical pretreatments involve associating anthocyanins
colorant but are susceptible to temperature-induced instability. A study
with copigments, such as flavonoids, alkaloids, amino acids, organic
of Thuy, Ben, Ngoc, and Tai (2022) found that the Vietnamese

Fig. 7. Correlation analysis and matrix plot of EY, TAC, TPC and DPPH from BPF extract. r: correlation coefficient and p-value for a 95% confidence interval.

11
S. Sai-Ut et al. Food Chemistry: X 22 (2024) 101484

acids, metals, and phenolics, to protect them from nucleophilic addition Declaration of competing interest
of water, preventing the conversion of the flavylium ion into pseudobase
or colorless forms (Bakowska-Barczak, Kucharska, & Oszmianski, 2003). The authors declare that they have no known competing financial
Food product color is directly influenced by the concentration of natural interests or personal relationships that could have appeared to influence
colorants. These research findings have practical applications for food the work reported in this paper.
businesses, enabling the use of safe natural colorants in coconut milk
jelly for Thai traditional food items. This approach enhances product Data availability
appeal and allows better control over color degradation, aligning with
the loss of anthocyanin content during processing and storage. The re­ No data was used for the research described in the article.
sults emphasize the importance of carefully applying cooking conditions
to preserve the natural color of the dessert product. Acknowledgments

3.10. Correlation analysis We express our gratitude to the Faculty of Science, Burapha Uni­
versity, for generously providing the necessary infrastructure and in­
The correlation analysis conducted between various response pa­ strument facilities for conducting our research.
rameters provides valuable insights into the relationships among these
variables, and a matrix plot is shown in Fig. 7. The primary variables References
analyzed were EY, TAC, TPC, and DPPH. The correlation coefficient
between EY and TAC was 0.41661, indicating a moderate positive cor­ Ali, S., & Ahmed, D. (2023). Comparing DES-mediated ultrasound- and heat-assisted
extraction of bioactive metabolites from Saussurea lappa and optimization by RSM
relation. Although this correlation was not statistically significant (p = and validation studies. Green Analytical Chemistry, 7, Article 100080. https://doi.
0.1244), it suggests a tendency for higher extraction yields to be asso­ org/10.1016/j.greeac.2023.100080
ciated with higher anthocyanin content. However, the correlation be­ Amić, D., Davidović-Amić, D., & Trinajstić, N. (1993). Application of topological indices
to chromatographic data: Calculation of the retention indices of anthocyanins.
tween EY and DPPH antioxidant activity was − 0.16379, indicating a
Journal of Chromatography A, 653(1), 115–121.
weak negative relationship (p = 0.5669). This suggests that a higher Arruda, H. S., Silva, E. K., Pereira, G. A., Angolini, C. F. F., Eberlin, M. N.,
extraction yield does not necessarily correspond to higher antioxidant Meireles, M. A. A., & Pastore, G. M. (2019). Effects of high-intensity ultrasound
process parameters on the phenolic compounds recovery from araticum peel.
activity, potentially due to the presence of non-antioxidant compounds
Ultrasonics Sonochemistry, 50, 82–95. https://doi.org/10.1016/j.
in the extracts or the specific nature of the antioxidant compounds ultsonch.2018.09.002
present. A more substantial positive correlation was observed between Bakowska-Barczak, A. (2005). Acylated anthocyanins as stable, natural food colorants –
TPC and DPPH antioxidant activity (0.59644, p = 0.0172). This signif­ A review. Polish Journal Of Food And Nutrition Sciences, 14, 107–116.
Bakowska-Barczak, A., Kucharska, A. Z., & Oszmianski, J. (2003). The effects of heating,
icant correlation indicates that phenolic compounds play a crucial role UV irradiation, and storage on stability of the anthocyanin-polyphenol copigment
in the antioxidant activity of the extracts. The strong relationship un­ complex. Food Chemistry, 81(3), 349–355.
derscores the importance of phenolics as key contributors to the overall Baskaran, A., Mudalib, S. K. A., & Izirwan, I. (2019). Optimization of aqueous extraction
of blue dye from butterfly pea flower. Journal of Physics, 1358, Article 012001.
antioxidant capacity of the extracts, validating the use of TPC as an in­ https://doi.org/10.1088/1742-6596/1358/1/012001
dicator of antioxidant potential in BPF extracts. These findings provide a Benavente-Garcı́a, O., Castillo, J., Lorente, J., Ortuño, A., & Del Rio, J. A. (2000).
better understanding of the relationships between the extracted com­ Antioxidant activity of phenolics extracted from Olea europaea L. leaves. Food
Chemistry, 68(4), 457–462. https://doi.org/10.1016/S0308-8146(99)00221-6
pounds and their bioactivities, informing future optimization and utili­ Benhanifia, M., Mohamed, W. M., Bellik, Y., & Benbarek, H. (2013). Antimicrobial and
zation of butterfly pea flower extracts in food applications. antioxidant activities of different propolis samples from North-Western Algeria.
International Journal of Food Science & Technology, 48, 2521–2527.
Berkani, F., Serralheiro, M. L., Dahmoune, F., Ressaissi, A., Kadri, N., & Remini, H.
4. Conclusions
(2020). Ultrasound assisted extraction of phenolic compounds from a jujube by-
product with valuable bioactivities. Processing, 8(11), 1441. https://doi.org/
This study aimed optimal conditions for extracting colorants from 10.3390/pr8111441
Brand-Williams, W., Cuvelier, M. E., & Berset, C. (1995). Use of a free radical method to
BPFs, revealing positive effects of high solid-to-liquid ratio on yield,
evaluate antioxidant activity. LWT - Food Science and Technology, 28(1), 25–30.
phenolic compounds, and antioxidants, while ultrasonic extraction time https://doi.org/10.1016/S0023-6438(95)80008-5
increased anthocyanin content, and ethanol concentration enhanced Campbell, S., Pearson, B., & Marble, S. (2019). Butterfly pea (Clitoria ternatea) flower
antioxidant activity. The statistical analysis indicated that the linear and extract (BPFE) and its use as a pH-dependent natural colorant. EDIS, 2019, 1–5.
https://doi.org/10.32473/edis-ep573-2019
quadratic terms of these variables significantly influenced the increased Chakraborty, S., Uppaluri, R., & Das, C. (2020). Optimization of ultrasound-assisted
pigment extraction from BPF, achieved through ultrasound treatment extraction (UAE) process for the recovery of bioactive compounds from bitter gourd
using the BBD method. The optimal conditions, as determined by the using response surface methodology (RSM). Food and Bioproducts Processing, 120,
114–122. https://doi.org/10.1016/j.fbp.2020.01.003
quadratic polynomial equation, include a 1:40 (g:mL) of sample-to- Charurungsipong, P., Tangduangdee, C., Amornraksa, S., Asavasanti, S., & Lin, J. (2020).
solvent ratio, 37 min of extraction time, and 61% ethanol concentra­ Improvement of anthocyanin stability in butterfly pea flower extract by co-
tion. The model’s reliability was confirmed through close agreement pigmentation with catechin. E3S Web of Conferences, 141, Article 03008. https://doi.
org/10.1051/e3sconf/202014103008
between predicted and actual values. BPF extract showed a potential as a Chemat, F., Rombaut, N., Sicaire, A. G., Meullemiestre, A., Fabiano-Tixier, A. S., & Abert-
natural and appealing food coloring agent through incorporation into a Vian, M. (2017). Ultrasound assisted extraction of food and natural products.
coconut milk jelly product. The optimized conditions could be applied Mechanisms, techniques, combinations, protocols and applications. A review.
Ultrasonics Sonochemistry, 34, 540–560. https://doi.org/10.1016/j.
for producing colorants used in the Thai dessert industry.
ultsonch.2016.06.035
Chen, L., Yang, M., Mou, H., & Kong, Q. (2018). Ultrasound-assisted extraction and
CRediT authorship contribution statement characterization of anthocyanins from purple corn bran. Journal of Food Processing
and Preservation, 42, Article e13377. https://doi.org/10.1111/jfpp.13377
Chusak, C., Ying, J. A. Y., Zhien, J. L., Pasukamonset, P., Henry, C. J., Ngamukote, S., &
Samart Sai-Ut: Writing – original draft, Visualization, Supervision, Adisakwattana, S. (2019). Impact of Clitoria ternatea (butterfly pea) flower on in
Conceptualization. Apisara Teksee: Validation, Methodology, Formal vitro starch digestibility,texture and sensory attributes of cooked rice using domestic
analysis. Jaksuma Pongsetkul: Writing – review & editing. Sirima cooking methods. Food Chemistry, 295, 646–652. https://doi.org/10.1016/j.
foodchem.2019.05.157
Sinthusamran: Writing – review & editing. Saroat Rawdkuen: Writing Coutinho, I. B., Freitas, A., & Maçanita, A. L. (2015). Effect of water content on the
– review & editing, Supervision. acid–base equilibrium of cyanidin-3-glucoside. Food Chemistry, 172, 476–480.
da Rocha, C. B., & Noreña, C. P. Z. (2020). Microwave-assisted extraction and
ultrasound-assisted extraction of bioactive compounds from grape pomace.
International Journal of Food Engineering, 16(1–2), 1–10. https://doi.org/10.1515/
ijfe-2019-0191

12
S. Sai-Ut et al. Food Chemistry: X 22 (2024) 101484

Dahmoune, F., Boulekbache, L., Moussi, K., Aoun, O., Spigno, G., & Madani, K. (2013). Lee, J., Durst, R. W., & Wrolstad, R. E. (2005). Determination of total monomeric
Valorization of Citrus Limon residues for the recovery of antioxidants: Evaluation anthocyanin pigment content of fruit juices, beverages, natural colorants, and wines
and optimization of microwave and ultrasound application to solvent extraction. by the pH differential method: Collaborative study. Journal of AOAC International,
Industrial Crops and Products, 50, 77–87. 88, 1269–1278.
Dahmoune, F., Spigno, G., Moussi, K., Remini, H., Cherbal, A., & Madani, K. (2014). Leong, T., Juliano, P., & Knoerzer, K. (2017). Advances in ultrasonic and megasonic
Pistacia lentiscus leaves as a source of phenolic compounds: Microwave-assisted processing of foods. Food Engineering Reviews, 9, 237–256. https://doi.org/10.1007/
extraction optimized and compared with ultrasound-assisted and conventional s12393-017-9167-5
solvent extraction. Industrial Crops and Products, 61, 31–40. Li, Q., Zhang, F., Wang, Z., Feng, Y., & Han, Y. (2023). Advances in the preparation,
Esmaeelian, M., Jahani, M., Feizy, J., & Einafshar, S. (2020). Effects of ultrasound- stability, metabolism, and physiological roles of anthocyanins: A review. Foods, 12
assisted and direct solvent extraction methods on the antioxidant and antibacterial (21), 3969. https://doi.org/10.3390/foods12213969
properties of saffron (Crocus sativus L.) corm extract. Food Analytical Methods, 1–14. Linares, G., & Rojas, M. L. (2022). Ultrasound-assisted extraction of natural pigments
Ferarsa, S., Zhang, W., Moulai-Mostefa, N., Ding, L., Jaffrin, M. Y., & Grimi, N. (2018). from food processing by-products: A review. Frontiers in Nutrition, 9, Article 891462.
Recovery of anthocyanins and other phenolic compounds from purple eggplant peels https://doi.org/10.3389/fnut.2022.891462
and pulps using ultrasonic-assisted extraction. Food and Bioproducts Processing, 109, Melgar, B., Dias, M. I., Barros, L., Ferreira, I. C. F. R., Rodriguez-Lopez, A. D., & Garcia-
19–28. https://doi.org/10.1016/j.fbp.2018.02.006 Castello, E. M. (2019). Ultrasound and microwave assisted extraction of Opuntia
Fossen, T., Cabrita, L., & Andersen, O. M. (1998). Colour and stability of pure fruit peels biocompounds: Optimization and comparison using RSM-CCD. Molecules
anthocyanins influenced by pH including the alkaline region. Food Chemistry, 63(4), (Basel, Switzerland), 24(19), 3618. https://doi.org/10.3390/molecules24193618
435–440. More, P. R., & Arya, S. S. (2021). Intensification of bio-actives extraction from
Gadioli Tarone, A., Keven Silva, E., de Freitas, D., Queiroz Barros, H., Baú Betim pomegranate peel using pulsed ultrasound: Effect of factors, correlation,
Cazarin, C., & Roberto Marostica Junior, M. (2021). High-intensity ultrasound- optimization and antioxidant bioactivities. Ultrasonics Sonochemistry, 72, Article
assisted recovery of anthocyanins from jabuticaba by-products using green solvents: 105423. https://doi.org/10.1016/j.ultsonch.2020.105423
Effects of ultrasound intensity and solvent composition on the extraction of phenolic Mori, K., Goto-Yamamoto, N., Kitayama, M., & Hashizume, K. (2007). Loss of
compounds. Food Research International, 140, Article 110048. https://doi.org/ anthocyanins in red-wine grape under high temperature. Journal of Experimental
10.1016/j.foodres.2020.110048 Botany, 58(8), 1935–1945. https://doi.org/10.1093/jxb/erm055
Ghosh, S., Sarkar, T., Das, A., & Chakraborty, R. (2021). Micro and nanoencapsulation of Multisona, R., Shirodkar, S., Arnold, M., & Gramza Michalowska, A. (2023). Clitoria
natural colors: A holistic view. Applied Biochemistry and Biotechnology, 193. https:// ternatea flower and its bioactive compounds: Potential use as microencapsulated
doi.org/10.1007/s12010-021-03631-8, 03631–8. ingredient for functional foods. Applied Sciences, 13, 2134. https://doi.org/10.3390/
Ghosh, S., Sarkar, T., Das, A., & Chakraborty, R. (2022). Natural colorants from plant app13042134
pigments and their encapsulation: An emerging window for the food industry. LWT - Olivas-Aguirre, F. J., Rodrigo-García, J., Martínez-Ruiz, N. D., Cárdenas-Robles, A. I.,
Food Science and Technology, 153, Article 112527. https://doi.org/10.1016/j. Mendoza-Díaz, S. O., Álvarez-Parrilla, E., González-Aguilar, G. A., de la Rosa, L. A.,
lwt.2021.112527 Ramos-Jiménez, A., & Wall-Medrano, A. (2016). Cyanidin-3-O-glucoside: Physical-
Giusti, M., & Wrolstad, R. (2003). Acylated anthocyanins from edible sources and their chemistry, foodomics and health effects. Molecules, 21(9), 1264. https://doi.org/
applications in food systems. Biochemical Engineering Journal, 14, 217–225. https:// 10.3390/molecules21091264
doi.org/10.1016/S1369-703X(02)00221-8 Patras, A., Brunton, N. P., & O’Donnell, C. (2010). Effect of thermal processing on
Goula, A. M., Ververi, M., Adamopoulou, A., & Kaderides, K. (2017). Green anthocyanin stability in foods; mechanisms and kinetics of degradation. Trends in
ultrasoundassisted extraction of carotenoids from pomegranate wastes using Food Science & Technology, 21(1), 3–11.
vegetable oils. Ultrasonics Sonochemistry, 34, 821–830. https://doi.org/10.1016/j. Plazzotta, S., Ibarz, R., Manzocco, L., & Martín-Belloso, O. (2020). Optimizing the
ultsonch.2016.07. 022 antioxidant biocompound recovery from peach waste extraction assisted by
Hasanah, N. N., Mohamad Azman, E., Rozzamri, A., Zainal Abedin, N. H., & Ismail- ultrasounds or microwaves. Ultrasonics Sonochemistry, 63, Article 104954. https://
Fitry, M. R. (2023). A systematic review of butterfly pea flower (Clitoria ternatea L.): doi.org/10.1016/j.ultsonch.2019.104954
Extraction and application as a food freshness pH-Indicator for polymer-based Soria, A. C., & Villamiel, M. (2010). Effect of ultrasound on the technological properties
intelligent packaging. Polymers, 15(11), 2541. https://doi.org/10.3390/ and bioactivity of food. Trends in Food Science and Technology, 21, 323–331. https://
polym15112541 doi.org/10.1016/j.tifs.2010.04.003
Houghton, A., Appelhagen, I., & Martin, C. (2021). Natural blues: Structure meets Surh, J., & Koh, E. (2014). Effects of four different cooking methods on anthocyanins,
function in anthocyanins. Plants, 10, 726. https://doi.org/10.3390/plants10040726 total phenolics and antioxidant activity of black rice. Journal of the Science of Food
Jeyaraj, E. J., Lim, Y. Y., & Choo, W. S. (2022). Antioxidant, cytotoxic, and antibacterial and Agriculture, 94, 3296–3304.
activities of Clitoria ternatea flower extracts and anthocyanin-rich fraction. Scientific Swain, T., & Hillis, W. E. (1959). The phenolic constituents of Prunus domestica 1. The
Reports, 12(1), Article 14890. https://doi.org/10.1038/s41598-022-19146-z quantitative analysis of phenolic constituents. Journal of the Science of Food and
Jokioja, J., Yang, B., & Linderborg, K. M. (2021). Acylated anthocyanins: A review on Agriculture, 10, 63–68.
their bioavailability and effects on postprandial carbohydrate metabolism and Thuy, N. M., Ben, T. C., Ngoc, P. T. B., & Tai, N. V. (2022). Application of butterfly pea
inflammation. Comprehensive Reviews in Food Science and Food Safety, 20, flower extract in processing some Vietnamese traditional foods. Journal Of Applied
5570–5615. https://doi.org/10.1111/1541-4337.12836 Biology & Biotechnology, 10(5), 143–150. https://doi.org/10.7324/
Kentish, S. E. (2017). Chapter 1 - engineering principles of ultrasound technology. In JABB.2022.100518
D. Bermudez-Aguirre (Ed.), Ultrasound: Advances for food processing and preservation Umair, M., Jabbar, S., Nasiru, M. M., Lu, Z., Zhang, J., & Abid, M. (2021). Ultrasound-
(pp. 1–13). Amsterdam: Elsevier. https://doi.org/10.1016/b978-0-12-804581- assisted extraction of carotenoids from carrot pomace and their optimization
7.00001-4. through response surface methodology. Molecules, 26, 6763. https://doi.org/
Khoo, H. E., Azlan, A., Tang, S. T., & Lim, S. M. (2017). Anthocyanidins and 10.3390/molecules26226763
anthocyanins: Colored pigments as food, pharmaceutical ingredients, and the Vidana Gamage, G. C., Lim, Y. Y., & Choo, W. S. (2021). Anthocyanins from Clitoria
potential health benefits. Food & Nutrition Research, 61(1), Article 1361779. https:// ternatea flower: Biosynthesis, extraction, stability, antioxidant activity, and
doi.org/10.1080/16546628.2017.1361779 applications. Frontiers in Plant Science, 12, Article 792303. https://doi.org/10.3389/
Kongpichitchoke, T., Hsu, J. L., & Huang, T. C. (2015). Number of hydroxyl groups on the fpls.2021.792303
b-ring of flavonoids affects their antioxidant activity and interaction with phorbol Wiyantoko, B. (2020). Butterfly pea (Clitoria ternatea L.) extract as indicator of acid-base
ester binding site of PKCδ C1B domain: In Vitro and in Silico Studies. Journal of titration. Indonesian Journal of Chemistry, 3, 22–32.
Agricultural and Food Chemistry, 63(18), 4580–4586. https://doi.org/10.1021/acs. Zafra-Rojas, Q. Y., González-Martínez, B. E., Cruz-Cansino, N. D. S., López-
jafc.5b00312 Cabanillas, M., Suárez-Jacobo, Á., & Cervantes-Elizarrarás, A. (2020). Effect of
Koraqi, H., Petkoska, A. T., Khalid, W., Kumar, N., & Pareek, S. (2023). Optimization of ultrasound on in vitro bioaccessibility of phenolic compounds and antioxidant
experimental conditions for bioactive compounds recovery from raspberry fruits capacity of blackberry (Rubus fruticosus) Residues cv. Tupy. Plant Foods for Human
(Rubus idaeus L.) by using combinations of ultrasound-assisted extraction and deep Nutrition, 75, 608–613. https://doi.org/10.1007/s11130-020-00855-7
eutectic solvents. Applied. Food Research, 3, Article 100346. https://doi.org/ Zhang, L., Fan, G., Khan, M. A., Yan, Z., & Beta, T. (2020). Ultrasonic-assisted enzymatic
10.1016/j.afres.2023.100346 extraction and identification of anthocyanin components from mulberry wine
Kumar, K., Srivastav, S., & Sharanagat, V. (2020). Ultrasound assisted extraction (UAE) residues. Food Chemistry, 323, Article 126714. https://doi.org/10.1016/j.
of bioactive compounds from fruit and vegetable processing by-products: A review. foodchem.2020.126714
Ultrasonics Sonochemistry, 70. https://doi.org/10.1016/j.ultsonch.2020.105325

13